. ,·· . ' .. . -(·
. ' :- .. ·, ..
· THE cHROMATOGRAPHic DETERMINATION :oF-· ms.lJuiJM TOXINS IN
~IZE.AssociATED-· ·. . . ·. . WITH. HtJMAN- . oESOPHAGEAl.:. . . cANcER.
by
ERIC WiLLIAM.· SYDENHAM I. I I: l
. \· .. ·.Thesis submitted fo·r the Degree
. . . .. ·MAs'l'ER OF SCIENCE . IN. ANALYTicAL . CHEMISTRY ·.I
'-· University ofat Capethe Town '1.' 1 i l I· University of Cape Town .. ' '
I September 19·89 . I I
. I. ,· ).
The copyright of this thesis vests in the author. No quotation from it or information derived from it is to be published without full acknowledgement of the source. The thesis is to be used for private study or non- commercial research purposes only.
Published by the University of Cape Town (UCT) in terms of the non-exclusive license granted to UCT by the author.
University of Cape Town ii-
. >
) . .
/- i -
·oedicated to my mother, the memory of my late father and especially to Gill and Shaun for their continuous love and encouragement iii
ACKNOWLEDGEMENTS
I am indebted to the following persons and institutions . who made the presentation of this thesis possible.
Dr Pieter G. Thiel for his continued interest, encouragement and guidance that have extended beyond the boundary of this study.
Dr Klaus R. Koch for his invaluable support and guidance du during the course of this study.
Prof Walter F.O. Marasas and Dr Dirk J. Van Schalkwyk for
their additional expertise with regard to the generation of mycological and statistical data.
Miss Sonja Stockenstr6m, Miss Liana Van der Westhuizen and
Dr Iain M. Moodie, for their advice and technical assistance.
Mrs Pauline Smith for her patience and excellent typing of this thesis.
Mrs Margie Walter for her excellent preparation of the many figures included in this thesis.
Mr Gordon Grierson for his assistance with regard to the ge neration of mass spectral data.
The South African Medical Research Council who made it possi- ble for me to pursue this study. iv
ABBREVIATIONS
AOAC Association of Official Analytical Chemists ATA Alimentary toxic aleukia DON Deoxynivalenol ECD Electron capture detector ELISA Enzyme-linked immunosorbent assay
FB1 Fumonisin B1 FB Fumonisin B 2 2 FID Flame ionization detector GC Gas chromatography GC/MS Gas chromatography - mass spectrometry GGT+ Gamma-glutamyl-transpeptidase-positive HPLC High performance liquid chromatography IR Infra-red LEM Equine leukoencephalomalacia MON Moniliformin MS Mass spectrum MRC south African Medical Research council NIV Nivalenol NMR Nuclear magnetic resonance RIA Radio-immunoassay TCA Tricarballylic acid TLC Thin layer chromatography uv Ultra-violet ZEA Zearalenone v
CONTENTS
ACKNOWLEDGEMENTS iii
ABBREVIATIONS iv
CHAPTER 1
INTRODUCTION
1.1 AIM OF THE STUDY 1
1.2 BACKGROUND 2
1.3 FUNGI AND FUSARIUM SPECIES IN PARTICULAR: 3 CLASSIFICATION
1.4 HUMAN AND ANIMAL MYCOTOXICOSIS 5
1.5 FUSARIUM GRAMINBARUM SCHWABE 8
1.5.1 Incidence and Distribution 8
1.5.2 Association of F. graminearum with human 8 diseases
1.5.3 Association of F. graminearum with animal 9 diseases
1.5.4 Mycotoxins produced by F. graminearum 10
1.6 FUSARIUM MONILIFORMB SHELDON 11
1.6.1 History and Distribution 11 Vi
1.6.2 Association of F. moniliforme with human 12 diseases
1.6.3 Association of F. moniliforme with animal 12 diseases
1.6.4 Mycotoxins produced by F. moniliforme 14
1.7 FUSARIUM SUBGWTINANS (WOLLENWEBER & 14
REINKING) NELSON, TOUSSOUN AND MARASAS
1.7.1 Incidence and Distribution 14 1.7.2 Association of F. subglutinans with human 14 diseases
1. 7.3 Mycotoxins produced,by F~ subglutinans 15
1.8 MYCOTOXINS RELEVANT TO THIS STUDY 15
1.8.1 Selected mycotoxins 15 1.8.2 The trichothecenes - introduction 16
1.8.3 The trichothecenes - structural classification 16 1.8.4 The trichothecenes - physico-chemical properties 19 1.8.5 Zearalenone - introduction and structure 20
1.8.6 Zearalenone - physico-chemical characteristics 21 1.8.7 Moniliformin - introduction and structure 21
1.8.8 Moniliformin - physico-chemical characteristics 22
1.8.9 Fumonisins (B1 and B2 ) - introduction and 22 structure
1.8.10 Fumonisin B1 - physico-chemical characteristics 23
1. 9 FUSARIUM MYCOTOXIN ANALYTICAL METHODOLOGY AND 24 OBJECTIVES OF THE STUDY vii
CHAPTER 2
EXPERIMENTAL
2.1 SAMPLE MATERIALS 28
2.1.1 Control maize 28
2.1.2 Fungal cultures -28 2.1.3 Transkeian maize samples - evaluation of 29
fumonisin B1 methodology 2.1.4 Transkeian maize samples - collection and 29 mycology
2.2 ANALYTICAL STANDARDS 30
2.3 EVALUATION OF EXISTING METHODOLOGY 31
2.3.1 General overview 31 2.3.2 Analytical methodology - type A trichothecenes 31 2 .. 3. 2 .1 Sample preparation - (Cohen and Lapointe, 31 1984) 2.3.2.2 Sample preparation - (Sydenham and Thiel, 32 1987)
2.3.3 Analytical standards - diacetoxyscirpenol and 33 T-2 toxin 2.3.4 Derivatization procedure - type A 34 trichothecenes 2.3.5 Chromatographic separation - type A 35 trichothecenes
2.4 ANALYTICAL METHODOLOGY - TYPE B TRICHOTHECENES 35
2.4.1.1 Sample preparation- (Tanaka et al., 1985a) 35 viii
2.4.1.2 Sample prepartion- (Scott et al., 1986) 37
.2.4.2 Analytical standards - nivalenol and 38 deoxynivalenol
2.4.3 Derivatization procedure - nivalenol and 38 deoxynivalenol
2.4.4 Chromatographic separation - nivalenol and 40 deoxynivalenol
2.5 ANALYTICAL METHODOLOGY - ZEARALENONE 40
2.5.1.1 Sample preparation - (AOAC, 1984) 40
2.5.1.2 Sample preparation - (Chang and DeVries, 41
1984)
2.5.1.3 Sample preparation- (Bagneris et al., 1986) 41
2.5.2 Analytical standard - zearalenone 42
2.5.3 Chromatographic separation - zearalenone 43
2.6 ANALYTICAL METHODOLOGY - MONILIFORMIN 44
2.6.1.1 Sample preparation- (Thiel et al., 1982b) 44
2.6.1.2 Sample preparation - (Scott and Lawrence, .44
1987)
2.6.2 Analytical standard - moniliformin 45 2.6.3 Chromatographic separation - moniliformin 46
2.7 INNOVATIVE METHODOLOGY 46
2.7.1 Sample preparation - fumonisin B1 46 2.7.2 Derivatization for HPLC separation 47
2.7.3 Hydrolysis, derivatization and GC separation 47
2.7.4 Secondary extraction and derivatization - 48
fumonisin B1 ix
2.7.5 Analytical standard - fumonisin a1 49 2.7.6 Chromatographic conditions - fumonisin B 1 50 (as its maleyl derivative) by HPLC
2.7.7 Chromatographic conditions - fumonisin a1 51 (as its fluorescamine derivative) by HPLC
2.7.8 Chromatographic conditions - TCA by capillary 52
GC
2.8 CONFIRMATORY TECHNIQUES 53
2.8.1 Improvement of chromatographic data 53
2.8.2 Chromatographic confirmation of nivalenol 53 and deoxynivalenol
2.8.3 Chromatographic confirmation of zearalenone 55
2.8.4 Chromatographic confirmation of moniliformin 55
2.8.5 Chromatographic confirmation of fumonisin Bl 57 2.8.6 Chromatographic confirmation of TCA 57
2.9 STATISTICAL ANALYSES 57
CHAPTER 3
RESULTS AND DISCUSSION
A: STANDARDS AND EXISTING METHODOLOGY
3.1 ANALYTICAL STANDARDS 59
3.1.1 Importance of authentic standards 59
3.1.2 Melting points 60
3.1.3 Thin layer and high performance liquid 61 chromatography X
3.1.4 Capillary GC analyses of HFBI and Tri-Sil TBT 62 derivatives 3.1.5 Infra-red absorption spectra 62 3.1.6 Ultra-violet absorptions spectra 63 3.1.7 Fluorescence spectroscopy 65 3.1.8 Nuclear magnetic resonance and mass 68 spectrometry techniques 3.1.9 Analytical standards - conclusion 68
3.2 COMPARISON OF TYPE A TRICHOTHECENE METHODOLOGY 68
3.2.1 Extraction procedures 68 3.2.2 Chromatographic procedures 69 3.2.3 Toxin recoveries and method detection limit 72 3.2.4 Type A trichothecene methodology - summary 74
3.3 COMPARISON OF TYPE B TRICHOTHECENE METHODOLOGY 7 4
3.3.1 Extraction procedures 74 3.3.2 Chromatographic procedures 75 3.3.3 Toxin recoveries, reproduciblity and method 78 detection limit 3.3.4 Type B trichothecenes - summary 79
3.4 COMPARISON OF ZEARALENONE METHODOLOGY 81
3.4.1 Extraction procedures 81 3.4.2 Chromatographic procedures 82 3.4.3 Toxin recoveries and method detection limit 85
3.4.4 Zearalenone methodology - summary 85
3.5 COMPARISON OF MONILIFORMIN METHODOLOGY 87
3.5.1 Extraction procedures 87 xi
3.5.2 Chromatographic procedures 88
3.5.3 Toxin recoveries and method detection limit 90
3.5.4 Moniliformin methodology - summary 91
B : INNOVATIVE METHODOLOGY
3.6 TLC DETERMINATION OF THE FUMONISINS 92
3.7 HPLC DETERMINATION OF FUMONISIN B1 (AS ITS 92 MALEYL DERIVATIVE
3.8 HPLC DETERMINATION OF FUMONISIN a1 (AS ITS 99 FLUORESCAMINE DERIVATIVE)
3.9 CAPILLARY GC OF FUMONISIN B1 103
3.10 SUMMARY OF FUMONISIN B1 METHODOLOGY . 107
CHAPTER 4
APPLICATION OF FUSARIUM MYCOTOXIN METHODOLOGY TO A SERIES OF HOME-GROWN TRANSKEIAN MAIZE SAMPLES ASSOCIATED WITH HUMAN OESOPHAGEAL CANCER RISK
4.1 OESOPHAGEAL CANCER IN THE TRANSKEI 109
4.2 FUSARIUM MYCOTOXIN ANALYSES 111
4.3 CHROMATOGRAPHIC ANALYSES AND CONFIRMATION 115
4.3.1 Type A trichothecene analyses 115
4.3.2 Type B trichothecene analyses 115
4.3.3 Moniliformin analyses 120
4.3.4 Zearalenone analyses 123
4.3.5 Tricarballylic acid analyses 126 xii
4.4. STATISTICAL ANALYSES OF THE CHEMICAL AND 132 MYCOLOGICAL RESULTS
4.4.1 overview 132
4.4.2 Moniliformin and Fusarium subglutinans 132
4.4.3 Zearalenone, nivalenol, deoxynivalenol and 134 Fusarium graminearum
4.4.4 Tricarballylic acid and Fusarium moniliforme 135
4.5 SUMMARY OF STATISTICAL ANALYSES 136
SUMMARY 141
REFERENCES 143 1
"• CHAPTER 1
IRTRODUCTIOH
1.1 AIM OF THE STUDY
Interest in the role played by a number of food-borne myco toxins in the etiology of ani.mal and human diseases and/or syndromes, has intensified over the last 25 years. It is known that these mycotoxins are as diverse in their chemical natu:J;:"e as are the fungi tha.t p~oduce them.
Lon9 before the Christian era food was subject to legislation and inspection. An example of early food 'legislation' was the law, promulgated by the King of the Hittites (Schuller ~t al., 1983), the first part of which reads "Thou shalt not taint the fat or the bread of thy neighbour, neither shalt thou bewitch the fat or the bread of thy neighbour". These days, the food laws not only.prohibit the introduction, deli very for introduc.tion, or receipt in commerce of adul te rated and misbranded food, but often includes specific legis lation that imposes limits or tolerances on the concentra- tions ·of specific contaminants in food. These contaminants can be of industrial or natural origin. The importance of mycotoxins to animal and huma.n he~l th might then be reflected ' . in ·the fact that of the natural contaminants, the mycotoxins are the most recent to be considered for legislative regula- tion.
"• 2
The necessity to obtain accurate and reliable data pertaining to the range and/or levels of mycotoxin contamination in a variety of food and feed substrates, intended for human or animal consumption, has therefore, become important. An inte gral part of this study will 'be the provision of the most suitable analytical methods for the determination of selected
Fusarium mycotoxins, in maize. The culmination will be the application of those selected methodologies to a series of Transkeian maize samples associated with human oesophageal cancer-risk.
1.2 BACKGROUND
Fungi or moulds are members of the lower plant species that do not contain chlorophyll and which reproduce by means of spores. Due to the absence of chlorophyll, they are unable to synthesize carbohydrates via the process of photosynthesis, as do the higher plant species. Consequently, fungi need to obtain their food by . growing either parasitically on the living tissues of plants, animals or man, or saprophytically on dead organic matter. It is due to the process by which they obtain their food that fungi in general pose a potential hazard to human health. During their growth stage, numerous fungi have the ability to produce a diverse range of secon dary metabolites (mycotoxins), which can be poisonous when ingested by animals or man (Marasas, 1988).
Studies concerning the death of 100,000 turkeys in Britain in 3
1960 from a disease first called "tuikey X disease" might be
regarded as the catalyst for modern research into mycotoxins
and their role in related diseases. Isolation and characteri~ sation of the causative agent for the "turkey X disease" led
to the discovery of the aflatoxins (Austwick, 1978), and the 'name of the disease was hence changed to "aflatoxicosis". The
impact of the discovery of "aflatoxicosis" as a distinct di sease syndrome, and the subsequent implication of aflatoxins as potential carcinogens in the human health chain (LeBreton
et al., 1962; Kraybill and Shimkin, 1964), had international implications. Although major research efforts centered on the aflatoxins, attention rapidly turned to other mycotoxins concerning their cause and effect relationship with a number
of idiopathic diseases. Hence, the worldwide concern for. "aflatqxins" and "aflatoxicosis" became a broader one for
"mycotoxins" and "mycotoxicosis".
1 • 3 FUNGI AND FUSARIUM SPECIES IN PARTICULAR:
CLASSIFICATION
Fungi and moulds have on the grounds of morphology been grou ped into specific genera. This study will deal exclusively with the genus Fusarium. Morphological differences have al-· so been used to divide the different spec1es within the genus Fusarium into subsections. The correct taxonomy and nomen clature are of importance, and since several classification systems have been used worldwide, discrepancies with regard to nomenclature of numerous species by different workers are 4
to be found in the literature. These differences have, there fore, led to a degree of confusion.
The genus Fusarium was first described by Link in 1808 (Ueno, 1983). In 1935, the genus was divided into 16 sections, 55 varieties and 22 forms (Wollenweber and Reinking, 1935). In 1945 this classification system was modified and the number of species was reduced to 9 (Snyder and Hansen, 1945). This system has been extensively used in several parts of the world. Toussoun and Nelson (Toussoun and Nelson, 1968) published an illustrated guide as an aid to the identifica tion of the fusaria based on the Snyder and Hansen classifi fication system.
Due to the confusion aroused within the literature, Nelson and co-workers (Nelson et al., 1983) published an illustrated guide to the identification of the genus, based predominantly on the classification system of Wollenweber and Reinking, (1935). However, the authors also incorporated certain characteristics of other classification systems. This attempt to establish a single universal system was augmented by the publication of a book .encompassing the identity and mycotoxicological data of the toxigenic Fusarium species (Marasas et al., 1984). In this study the system of Nelson et al., (1983) will be used when reference is made to any Fusarium species. 5
1.4 HUMAN AND ANIMAL MYCOTOXICOSIS
Since an integral part of this study concerns the prevalence of three specific Fusarium species (ie F. graminearum Schabe of the Section Discolor and F. moniliforme Sheldon and F. subglutinans (Wollenw. & Reinking) Nelson, Toussoun & Marasas of the Section Liseola), the mycotoxicological data pertai ning to these species will be dealt with separately in sec tions 1.5 to 1.7.
Mould induced food poisoning has posed a threat to human health for more than 2000 years. It has been suggested that an association exists between food-borne fungi and the plagues of Egypt, during biblical times (Schoental, 1984).
The oldest known outbreak of human mycotoxicosis is that of "ergotism". A disease occurred sporadically throughout cen tral Europe during the Middle Ages, which was characterised by a sensation of cold hands and feet which progressed to an intense burning sensation. In advanced cases the extremities of the victim became gangrenous and necrotic. Belief on the part of affected victims that relief from the intense burning sensations could be sought by visiting the shrine of St Anthony, resulted in the disease adopting the popular name "St Anthony's fire". The disease became associated with the ingestion of bread made from rye infected with the fungus Claviceps purpurea (Shank, 1978). The fungus produqed er gots, but it was only in the seventeenth century that 6
alkaloids produced by the ergots, were found to be responsible for the disease.
In the years 1942-1947 septic angina or alimentary toxic aleukia (ATA), a serious and often fatal disease, accompanied by extreme leukopenia, sepsis and the depletion of bone marrow, occurred widely in the Orenburg district of the USSR (Joffe, 1978). It is estimated that ATA claimed the lives of approximately 10% of the population of the region (correspon ding to hundreds of thousands of people), following the in gestion of overwintered grain infected with Fusarium poae Peck and Fusarium sporotrichioides Sherb. Initially, ATA was thought to be due to a deficiency of vitamins B1 and c and riboflavin. However, later work re~ulted in the isolatidn of the potent trichothecene T-2 toxin from cultures of F.poae and F. sporotrichioides isolated from the overwintered grains (Bamburg and Strong, 1971; Mirocha and Pathre, 1973). Using the isolated toxin, the characteristic symptoms of ATA were reproduced in cats (Lutsky et al., 1978), thereby substantiating the role of T-2 toxin in the etiology of ATA.
Cases of animal mycotoxicoses have been well documented (Christensen, 1979; Marasas et al., 1984). Many of these out breaks have been related to the consumption of feeds contami nated with a wide range of mycotoxins produced by various genera of toxigenic fungi. Distinct diseases or syndromes emanating from specific genera of fungi or group of myco toxins have been collectively named (ie Fusariotoxicosis is 7
the poisoning ·of animals resulting from the consumption of feed contaminated by toxin(s) produbed by fungi of the genus
Fusarium). Similarly~ aflatoxicosis and ochratoxicosis re- late to syndromes having their etiology linked to the presence of aflatoxins and ochratoxin, respectively.
A specific example of fusariotoxicosis is that of the hae- morrhagic syndrome (mouldy maize toxicosis) which is associa- ted with the consumption of mouldy cereals contaminated with
' F. sporotrichioides. outbreaks of the syndrome are characte- rised by bloody diarrhoea, necrotic oral lesions, haemorrha- gic gastro-enteritis and extensive haemorrhages in many organs in animals such as cattle, pigs and poultry (Marasas et al., 1984). The characteristic pathological changes have been reproduced experimentally with crystalline trichothe- cenes isolated from cultures of toxic strains of F. sporotri- chioides associated with outbreaks of the syndrome ·(Marasas
et al., 1984) 0 In the United states, Australia and New Zealand, sporadic outbreaks of a disease in cattle, known as "fescue foot", has been characterised by lameness, loss of weight, arched back, elevated body temperature and dry gangrene involving the hind .feet, tail tip and ears. Only one of the characteristic features of this disease (ie necrosis of the tip of the tail) has been reproduced experimentally by the administration of butenolide, a toxin known to be produced by F. sporotrichioides (Grove et al., 1970; Kosuri et al., 1970). However since butenolide has never been found to occur naturally in toxic fescue hay, the 8
causative role of butenolide and F. sporotrichioides in
fescue foot of cattle has not been proven (Marasas et al., 1984).
1 • 5 FUSARIUM GRAMINEARUM SCHWABE
1.5.1. Incidence and Distribution
Fusarium graminearum belongs to the Section Discolor (Nelson et al., 1983). It has a worldwide distribution as a pathogen that causes root-, foot-, crown-, stem-, and ear rot as well as head blight (scab) of cereals (Marasas et al., 1984). In
Australia, F. graminearum is the predominant Fusarium fun- gus associated with crown rot of wheat (Burgess et al., 1975). Two populations of the fungus have been diffe rentiated (Francis and Burgess, 1977) and designated as F. graminearum Group 1 and Group 2. Members of Group 1 are nor mally associated with diseases of the crowns of plants, whereas those members of Group 2 are associated with diseases of the aerial parts of the plants.
1.5.2. Association of F. graminearum with human diseases outbreaks of "scabby grain" intoxication or "Alkakabi-byo" in humans have been reported in both the USSR and Japan. In the USSR, victims suffered headaches, chills, nausea, vertigo, vomiting and visual abnormalities (Dounin,_ 1926), following 9
the ingestion of bread prepared from "scabby" rye infected with F. graminearum. Several outbreaks of the syndrome have occurred in Japan and Korea (Yoshizawa, 1983), where the · dis.ease is more commonly known as "Akakabi-byo". In two sepa rate incidents, a number of people in Hokkaido, Japan com plained of headaches, chills and abdominal pains amongst other symptoms, following the consumption of noodles prepared from red-mould contaminated flour (Kurata, 1978). F. g:rami- nearum was the predominant species isolated from Japanese barley, oats, rye and rice (Tsunoda, 1970; Yoshizawa and Morooka, 1977). A large scale cereal scab epidemic in Korea
(caused by F. graminearum), resulted in 80% loss·es in the barley yield (Yoshizawa, 1983). Similar clinical symptoms as . previously described resulted from the ingestion of food pre pared from the infected grain. The possibility that F. sporotrichioides was also involved in Akakabi-byo has been reported (Marasas et al., 1984). Although both nivalenol and deoxynivalenol (two type B trichothecenes amongst other mycotoxins known to be produced by F. graminearum and F. sporotrichioides) have frequently been reported to occur naturally in Japanese grains and grain products (Tanaka et al., 1985a, 1985b), neither of these potentially potent
' trichothecenes have been implicated in any case of human mycotoxicoses.
1.5.3. Association of F. graminearum with animal diseases
Many animals, particularly pigs, have refused to eat F. gra- 10
minearum infected grain, whilst those animals ingesting even
small amounts of the scabby grain were found to vomit. The active compoUnd was eventually isolated and characterised (Vesonder et al., 1973). The compound was given the trivial name. "vomitoxin", which is still used extensively, although it is more correctly termed ."deoxynivalenol".
Hyperestrogenism (vulvo-vaginitis), a disease normally asso ciated with pigs, is characterised by a number of clinical signs including pseudo-estrus and swollen vulvas in young gilts, prolapse of the vagina, enlargement of the mammary glands and infertility (Nelson et al., 1973; Aucock et al., 1980). In extreme cases death can result from septicemia fol lowing necrosis of the exposed prolapsed parts. McNutt and co-workers (McNutt et al., 1928), were the first to associate hyperestrogenism with the consumption of mouldy maize. The disease was induced with maize cultures of F. graminearum, isolated from fungally contaminated feed implicated in an outbreak of porcine hyperestrogenism ( stob et al.,. 1962). Using a bioassay technique the causative compound was isolated and given the trivial name F-2 toxin (Christensen et al., 1965). The chemical structure of F-2 toxin was determined (Urry et al., 1966) and the name of the toxin was changed to "zearalenone".
1.5.4. Mycotoxins produced by F. graminearum
The following mycotoxins have been reported to be produced by 11
F. graminearum in culture (Marasas et al., 1984):
butenolide, deoxynivalenol, deoxynivalenol monoacetate, deo xynivalenol diacetate, diacetoxyscirpenol, fusarenon-X, HT-2 toxin, monoacetoxyscirpenol, neosolaniol, nivalenol, T-2 toxin and zearalenone.
1.6 FUSARIUM HONILIFORME SHELDON
1.6.1. History and distribution
Fusarium moniliforme belongs to the section Liseola (Nelson et al., 1983) and is one of the most prevalent fungi associated with human and animal dietary staples, occurring on a wide variety of plant hosts. Saccardo (Saccardo, 1881) first described the fungus, found on Italian maize kernels,
naming it Oospora verticillioides. This fungus was implica ted as a possible cause of the disease "pellagra" due to it's prevalence in maize associated with the disease in Italy and
Russia (Tiraboschi, 1905; Von Deckenbach, 1907)~ Widespread outbreaks of an animal disease in the United States were re ported (Peters, 1904). The fungus most commonly associated with the maize consumed by the animals was identified as F. moniliforme Sheldon (Sheldon, 1904). It was later shown that F. moniliforme was identical to o. verticillioides (Manns and Adams, 1923). 12
1.6.2. Association of F. moniliforme with human diseases
In southern Africa, the highest human oesophageal cancer rate occurs in the south-western districts of the Transkei, while the rate is relatively low in the north-eastern region of the Transkei (Marasas et al., 1981, 1988a), where maize is the dietary staple in both areas. In a comparative study of the mycoflora of the local home-grown maize of the two areas, a significantly higher incidence of F. moniliforme was found to be present in the maize produced in the high-rate area of the Transkei (Marasas et al., 1981, 1988a). Several isolates of F. moniliforme taken from maize emanating from the high rate area have been found to be acutely toxic to ducklings. The acute and long term effects in experimental animals of two of these isolates (MRC 826 and MRC 602) have been determined and although these isolates have been shown to affect a variety of organs in different animal species (Marasas et al., 1984), no experimental evidence has been obtained to indicate that they can cause oesophageal cancer. In China, F. moniliforme is one of the most frequently encountered fungi associated with foodstuffs in the Lin Xian county in Henan province, which is one of the highest oesophageal cancer risk areas of the world (Li et al., 1979,
1980).
1.6.3. Association of F. moniliforme with animal diseases
Equine leukoencephalomalacia (LEM) is a neurotoxic disease of 13
horses, donkeys and mules. It is characterised by liquefac
tive necrotic lesions in the white matter of the cerebral hemispheres (Marasas et al., 1984). Mortalities in horses, normally preceded by marked nervous symptoms prior to death, have been recorded since 1850. The disease has been repro duced experimentally in horses on a number of occasions, by feeding naturally contaminated mouldy maize (Butler, 1902; Biester et al., 1940). It was only in 1971 that F. monili- forme was identified as the causative fungus (Wilson, 1971; Wilson and Maronpot, 1971). Culture material of F. monili forme MRC 826, isolated from maize intended for human con sumption in the Transkei, has also been shown to induce LEM in horses (Kriek et al., 1981). Clinical signs similar to those of equine LEM have also been reported in "bean hulls poisoning", a disease of horses which is prevalent in the Hokkaido district of Japan (Ueno et al., 1972). However, as yet it is not known whether this disease is identical to LEM.
Efforts to isolate and characterise the F. moniliforme my cotoxin responsible for LEM has been in progress for approxi mately 20 years. Using a bioassay technique as a monit6r for cancer-promoting principles, the fumonisins were isolated from culture material of F. moniliforme MRC 826 (Gelderblom et al., 1988). The major compound isolated was given the
11 trivial name "fumonisin B1 - (FB1 ). LEM was reproduced expe rimentally in a horse by the intravenous injection of FB1 isolated from MRC 826 (Marasas et al., 1988b). 14
1.6.4. Mycotoxins produced by F. moniliforme
The following mycotoxins have been reported to be produced by F. moniliforme in culture (Marasas et al., 1984, Gelderblom et al., 1988): fusaric acid, fusarin c, fusariocins, moniliformin, zearale none, fumonisin B , fumonisin B . 1 2
1. 7 FUSARIUM SUBGW'l'INANS (WOLLENWEBER & REINKING) NELSON,
TOUSSOUN AND MARASAS
1.7.1. Incidence and distribution
Fusarium subglutinans belongs to the Section Liseola (Nel son et al., 1983). Relatively little information is available on the incidence and distribution of F. subglutinans due to it's apparent misidentification as F. moniliforme {Booth, 1971). It has been suggested that F. subglutinans occurs on a wide range of hosts, in many areas of the world {Marasas et al., 1984). Booth (Booth, 1971) claimed that F. subglutinans had a lower optimum temperature fGr growth and that it would be more predominant in more temperate climates than F.moniliforme.
1.7.2. Association of F. subglutinans with human diseases
Fusarium subglutinans is one of the most prevalent fungi as sociated with home-grown maize produced in the high oeso- 15 phageal cancer risk areas of the Transkei (Marasas et al.,
1981). One.of the isolates of F. subglutinans from Trans keian maize was shown to be highly toxic to ducklings and rats and to produce large quantities of the Fusarium myco toxin moniliformin, in culture (Kriek et al., 1977). There is, however, no evidence that either .F. subglutinans or moniliformin are involved in the etiology of oesophageal cancer.
1.7.3. Mycotoxins produced by F. subglutinans
The following mycotoxins have been reported to be produced by F. subglutinans in culture (Marasas et al., 1984): fusaric acid, moniliformin
1 • 8 MYCOTOXIHS RELEVANT TO THIS STUDY
1.8.1. Selected mycotoxins
Although the fusaria in general, and the species outlined in sections 1.5 to 1.7 in particular, have been shown to pro duce a diverse range of secondary metabolites, only a few of these metabolites occur naturally (the majority occurring only in cultures). For the purpose of this study, the presence of 7 of these mycotoxins (deoxynivalenol, diacetoxy scirpenol, fumonisin B1 , moniliformin, nivalenol, T-2 toxin and zearalenone) will be taken into consideration. The chemical characteristics of each of these toxins will be 16 discussed in the following sections (1.8.2 to 1.8.10).
1.8.2. The trichothecenes - introduction
The trichothecenes are a chemically related group of royce toxins produced by various strains of Fusarium, Myrothe cium, Trichoderma and other genera of imperfect fungi. The first known trichothecene, trichothecin, was isolated from a culture of Trichothecium roseum Link (Freeman and Morrison, 1948). Phytotoxic metabolites were isolated from F. scirpi Lambotte & Fautr. (Brian et al., 1961), the major metabolite being named diacetoxyscirpenol, based on the fungal name and the presence of two acetyl groups present in the structure. Within several years, T-2 toxin, nivalenol and it's acetyla ted derivative, fusarenon-X, had been isolated (Bamburg et al., 1968, Morooka and Tatsuno, 1970). Although to date, in excess of 70 trichothecenes have been isolated and chemically characterised, the vast majority of these only occur in cultures under controlled laboratory conditions. Only a few of the trichothecenes (including nivalenol, deoxynivalenol, diacetoxyscirpenol and T-2 toxin) have been found to occur naturally on food and feedstuffs (Ueno, 1983; Kurata and Ueno, 1983), and it is these four trichothecenes which will be included in this study.
1.8.3. The trichothecenes - structural classification
The collective name for the structure of a tricyclic skeleton 17
composed of cyclohexane, cyclopentane, 6-membered oxyrane
rings and four methyl groups was proposed as "trichothecane" (Godtfredsen et al., 1967). However, most of the naturally occurring compounds contain an epoxide ring and a double bond
at c-a2,13 and C-9,10, respectively, and are therefore, cha- racterised by the 12,13-epoxytrichothec-9-ene ring system, hence the name "trichothecene", the basic structure of which is shown in Fig. 1.1.
H
'r.:::2-.-----::3+ -- R 1
I I RS I R4I
FIG. 1.1 Basic structure of the 12,13-epoxytrichothec-9-enes
Based on a number of chemical characteristics, the trichothe- . cenes have been divided into 4 distinct gioups (or types), namely A,B,C and D (Ueno, 1983). The four trichothecenes in- eluded in this study are members of groups A and B. Group A trichothecenes (diacetoxyscirpenol and T-2 toxin) lack a car bonyl at the C-8 position and have hydroxyl or acetyl groups 1 4 at R , R2 , R3 and R (Table 1.1). Group B trichothecenes 18
(deoxynivalenol and nivalenol) have a carbonyl at the C-8 position and mostly hydroxyls at R1 , R2 , R3 and R4 (Table 1.2).
TABLE 1.1 CHEMICAL STRUCTURES OF GROUP A TRICHOTHECENES INCLUDED IN THIS STUDY
)
H H
H
I I H
Trichothecene R1 R2 R3 R4 Rs Diacetoxyscirpenol OH OAc OAc H H Neosolaniol monoacetate OH OAc OAc H OAc T-2 toxin OH OAc OAc H Oiv
OAc =Acetate (-O-COCH3 ) Oiv = Isovalerate (-O-COCH2CH(CHJ) 2 ) 19
TABLE 1.2 CHEMICAL STRUCTURES OF GROUP B TRICHOTHECENES
INCLUDED IN THIS STUDY
Trichothecene
Nivalenol OH OH OH OH
Fusarenon-X OH OAc OH OH
Deoxynivalenol OH H OH OH
1.8.4. The trichothecenes - physico-chemical properties
The trichothecenes are generally colourless, crystalline com- pounds, which are soluble in moderately polar organic solvents such as ethanol, acetone, ethyl acetate and chloro- form, but only slightly soluble in water (Ueno, 1980). Due to the absence of any conjugat~d double bonds in their 20
structures, the group A trichothecenes do not exhibit UV
absorbance in either the ultraviolet or visible region of the spectrum, nor any fluorescence characteristics. Conversely, the presence of carbonyl group at the C-8 position present in
the structures of the group B trichothecenes results in an absorbance between 215-226 nm.
1.8.5. Zearalenone - introduction and structure
Although zearalenone is referred to as a mycotoxin it should be classified as a non-steroidal hormone as it is non-toxic and induces estrogenic activity in animals, especially pigs, at relatively low dietary levels (Christensen, 1979). The chemical structure of zearalenone was elucidated by Urry et al., (1966) as 6-(10-hydroxy-6-oxo-trans-1-undecenyl)-B-re- sorcyclic acid-~-lactone. Zearalenone exists in equilibrium between the cis and trans isomers (Fig. 1.2), however un- der ultraviolet irradiation the equilibrium is moved in the direction of the cis isomer.
OH 0 OH 0
HO HO
0
FIG. 1.2 Chemical structure of zearalenone isomers 21
Several derivatives of zearalenone have been reported, but the majority of these have only been detected in culture or after microsomal metabolism_by mammalian systems (Mirocha et al., 1977).
1.8.6. Zearalenone - physico-chemical characteristics
Zearalenone is an optically active, white, odourless crys o talline compound with a melting point range of 161-164 c, and a molecular mass of j18 (Pohland et al., 1982). The compound is soluble in ethanol, benzene and dichloromethane but only partially soluble in water. The ultraviolet spectrum exhibits absorption maxima at 236 nm, 274 nm and 316 nm (Pohland et al., 1982). Although zearalenone exhibits fluorescence cha- racteristics, some confusion exists in the literature concer- ning the corresponding fluorescence data (Trenholm et al., 1984; Bennett et al., 1985; Bagneris et al., 1986).
1.8.7. Moniliformin - introduction and structure
Moniliformin (Fig. 1.3) was first isolated from F. monili- forme and identified by x-ray crystallography as the potas- sium salt of 3-hydroxycyclobut-3-ene-1,2-dione (Cole et al., 1973; Springer et al., 1974). However the free acid had pre viously been chemically synthesized (Hoffman et al., 1971;
Pinske, 1972). 22
0~ ijo ~c c
cI c M - NaorK / H ' OM
FIG. 1.3 Chemical structure of moniliformin
1.8.8. Moniliformin - physico-chemical characteristics
Moniliformin is soluble in water and exhibits absorption maxima at 229 nm and 254 nm in the ultraviolet region of the spectrum (Cole and Cox, 1981). Isolated from a number of Fusarium species as the potassium, sodium, or a mixture of both salts, it decomposes without melting at temperatures up to 350°C.
1.8.9. Fumonisins (B1 and B2 ) - introduction and structure
Using a short-term cancer initiation-promotion bioassay with diethylnitrosamine-initiated rats and the induction of gamma- g1utamy1-transpeptidase-positive (GGT+) foci as endpoint, the fumonisins were fsolated from maize culture material of F. moniliforme MRC 826 (Gelderblom et al., 1988). Two distinct compounds were isolated and given the trivial names"fumoni- 23
II sin B1 and "fumonisin B2" (FB1 and FB 2 , respectively). The structure of the major component FB1 is given in Fig; 1.4.
FIG. 1.4 Chemical st~ucture of fumonisin B 1
The structures of four fumonisins were elucidated by Bezui denhout et al. (1988) using mass spectrometry and 1H and 13c N.M.R. spectroscopy. as the .diester of propane-1,2,3-
tricarboxylic acid and either 2-acetylamino or 2-amino-12,16~ dimethyl-3,5,10,14,15-pentahydroxycosane as well as in each case the c-10 deoxy analogue. In each case the C-14 and C-15 hydroxy groups are involved in the ester formation with the terminal carboxy group of propane-1,2,3-tricarboxylic acid (Tricarballylic acid- TCA).
1.8.10. Fumonisin B1 - physico-chemical characteristics
Fumonisin B1 has only recently been isolated, initially in an impure form (Gelderblom et al., 1988). It is soluble in water and has a molecular mass of 721. Indications are that FB1 24
most probably exists as the salt form in nature. It does not exhibit any absorbance or fluorescence characteristics in the ultraviolet or visible regions of the spectrum.
1. 9 FUSARIUM MYCOTOXIN ANALYTICAL METHODOLOGY AND
OBJECTIVES OF THE STUDY
A number of analytical methods have been developed and pub lished for the determination of single or groups of royce toxins in a number of different matrices. Although desirable, methods for the co-determination of several unrelated royce toxins (so called multi-mycotoxin methods), have been repor ted and reviewed (Steyn, 1981), but on the whole, these methods have been found to be impractical .. The majority of these methods have utilized numerous and often laborious pre chromatographic sample preparation procedures followed by thin layer chromatographic (TLC) separation. Due to the di versity of the rnycotoxins with regard to their respective chemical characteristics, these multi-mycotoxin methods have suffered from highly variable recoveries for the separate rnycotoxins.
Consequently, since the early 1980's, there has been a per ceptual shift in emphasis, away from the multi-mycotoxin approach towards the development of methods aimed at the de termination of singular or groups of closely related rnyco toxins, in food substrates. Several books dealing with myco toxin methodology have been published, a most recent and corn- 25
prehensive addition being that edited by Dr R.J. Cole (Cole, 1986).
• It is interesting to note that TLC is still used extensively for the estimation of mycotoxin contamination .. In a recently published survey (Van Egmond, 1989), 66 countries responded
to enquiries concerning existing or proposed legislati~n and methodology, pertaining to mycotoxin contamination of food. With regard to the aflatoxins· (the most regulated of the mycotoxins), 64% of the countries reported having regulations and official methods of analysis, while 36% of the countries
reported that regulations were either currently being
considered, proposed, or were in fact non-~xistant. Only 28% of those countries having official methods, however, used
·analytical techniques other than ~LC (Van Egmond, 1989).
The 1984 edition of the Official Methods of Analysis of the
Association of Official Analytical Chemists (AOAC~ 1984) lists predominantly TLC methods of analysis for the determination of each mycotoxin within the section concerning Natural. Poisons. The general referee for mycotoxins, Dr P.M. Scott, regularly reviews the literature pertaining to myco
toxin research including methodology. In the last thr~e reports (Scott, 1987, 1988, 1989), publications concerning mycotoxin methodology have included the increased use .of more sophisticated separation and detection techniques, such as high performance liquid chromatography, gas chromatography and mass spectrometry, and it is foreseen that these 26 techniques will be adopted by the AOAC, if not as replacement techniques for TLC analysis, then as confirmatory techniques. The use of radio-immunoassay (RIA) and enzyme-linked immunosorbent assay (ELISA) techniques, as screening procedures (predominantly for the analysis of the aflatoxins) also appear to have gained a great deal of attention (Park et al., 1989).
This investigation deals with the optimization of the condi- tions for the determination of 7 selected Fusarium myco- toxins in food, through the following objectives:
1. The screening and comparison of, where possible, exis
ting toxin ~ethodology.
2. Where necessary, the optimization of existing methode-· logy.
3. Where necessary, the development of innovative methodo logy.
4. The application of the most suitable methods for the de termination of the selected toxins in a series of home grown ·maize samples from different human oesophageal cancer risk areas of the Transkei, southern Africa.
5. Where possible, the verification of the presence or ab sence of the selected toxins determined in objective 27
number 4.
6. The statistical correlation between the toxin analytical
and previously generated mycological results, for those
samples covered in objective number 4.
An overview of the parameters covered in this investigation
is represented schematically in Figure 1.5.
GROUND MAIZE SAMPLE
PRIMARY EXTRACTION PROCEDURE
PURIFICATION PROCEDURES
DERIVATIZATION PROCEDURES (where necessary)
CHROMATOGRAPHIC SEPARATION PROCEDURES
CONFIRMATORY METI-IODS
FIG. 1.5 Schematic representation of investigation 28
CHAPTER 2
EXPERIMENTAL
2 • 1 SAMPLE MATERIALS
2:1.1 Control maize
First grade, yellow maize kernels were u~ed as a control ~a- trix to evaluate a number of the analytical methods' for the determination of the Fusarium mycotoxins listed in Sections 1.8.1-1.8.10. The kernels were dried at 50°C before being finely ground in a laboratory mill and the material was main tained at 5°C prior to analysis. Recovery experiments were conducted using this control maize, to which authentic myco toxin standards of known concentration were added . prior to extraction and analysis.
2.1.2 Fungal cultures
Samples of maize kernels on which selected fungi had been grown were also used for the evaluation of the analytical methods for the determination of the trichothecenes diace- toxyscirpenol and T-2 toxin, and for FB1 and FB2 . Lyophilized conidia of F. armeniacum (MRC 4481) and F. moniliforme
I (MRC 826) from the collection of the South African Medical Research Council (MRC), were used to inoculate autoclaved yellow maize kernels (400 g maize and 400 ml ~ater) in 2- 29
litre glass fruit jars. The cultures were incubated at 25°C in the dark for 21 days, after which the material was dried
{50°C for 24 hours) and finely ground in a laboratory mill. The dried fungal culture material was stored at s0 c prior to analysis.
2.1.3 Transkeian maize sample - evaluation of fumonisin B 1 methodology
A sample of mouldy maize ears of the 1978 crop was also used to evaluate the fumonisin B1 methodology. The sample was ob tained from a farm in the Butterworth district of Transkei, during July 1978 (Thiel et al., 1982). Visibly Fusarium in- fected ears were selected and hand shelled. The kernels (a mixture of Fusarium-infected, healthy, and kernels infec- ted with other fungi) were retained as sample M-84 {Thiel et al., 1982; Gelderblom et al., 1984). Two subsamples, one con- taining predominantly healthy kernels and the other predomi- nantly Fusarium-infected kernels, were selected from sample
M-84 and retained as samples M-84/C and M-84/F, respectively. Each sample was ground in a laboratory mill and stored at 5 0 c prior to analysis.
2.1.4 Transkeian maize samples - collection and mycology
The maize samples described in sections 2.1.1.-2.1.3, were used primarily for the evaluation of the Fusarium mycotoxin analytical methodology. The final phase of this study was the 30
application of the most suitable methods for each mycotoxin,
to a series of home-grown maize samples collected from areas of the Transkei.
Samplei of home-grown maize of the 1985 crop intended for human consumption were.collected from 12 households each in the Kentani and Bizana areas of the Transkei (Marasas et al.,
1988). The maize ears had previously been ~eparated and stored by the housewife of each household into "good" ears (intended for porridge making) and "mouldy" ears (for beer making and animal feed). The ears were hand shelled and the fungi isolated from 100 surface-sterilised kernels per sample as previously described (Marasas et al., 1981). Fusarium species were identified according to the system of Nelson and co-workers (Nelson et al., 1983). Kernels harvested from the "mouldy" maize ears were finely ground in a laboratory mill,
stored at 5°C and used for the purpose of this study~
· Mycological data pertaining to these samp~es will be presented in the results section for statistical purposes, although the mycological data itself was generated by Prof W.F.O. Marasas (Marasas et al., 1989).
2.2 ANALYTICAL STANDARDS
The analytical standards used in this investigation wer~ either obtained from commercial sources or were previously isolated within the Institute. The identity and purity of each standard was assessed using a number of techniques in- 31
eluding: thin layer chromatography (TLC), gas chromatography
(GC), high performance liquid chromatography (HPLC), melting point, infra-red (IR), ultra-violet (UV), fluorescence and nuclear magnetic resonance spectroscopy, and mass spectrome try (MS). The results generated were, where possible compared .with those previously published {Cole and Cox, 1981; Pohland et aL, 1982).
2.3 EVALUATION OF EXISTING METHODOLOGY
2.3.1 General overview
Several published methods for the determination of the Fusarium toxins previously mentioned in sections 1.8.1 1.8.10, in maize and other substrates, were evaluated. Sample preparation and chromatographic separation of these methods
will be dealt with separately in sections 2.3.2 - 2.6.3.
2.3.2 Analytical methodology - type A trichothecenes
Two methods for the simultaneous determination of diacetoxy scirpenol and T-2 toxin in maize were evaluated (Cohen and Lapointe, 1984; Sydenham and Thiel, 1987).
2.3.2.1 Sample preparation - (Cohen and Lapointe, 1984)
A subsample of ground maize (50 g) was extracted with metha nol/water (1:1, 250 ml) by blending at high speed for 5 32
minutes. The mixture was then centrifuged for 5 mins at 2000
rpm, followed by filtration. An aliquot of the filtrate (50 ml) was mixed with 30% aqueous ammonium sulphate (100 ml) and
Celite {20 g), stirred for 2 mins and filtered. The filtrate was partitioned with ethyl acetate {4 x 100 ml) which was then dried with anhydrous sodium sulphate, re-filtered and evaporated to dryness under vacuum. The residue, dissolved in dichloromethane (2 ml) was transferred to a silica gel Sep- pak cartridge, the flask being washed with further aliquots of dichloromethane (3 x 5 ml) which were quantitatively transferred to the same cartridge. The eluate (first 5 ml) ·from the cartridge was discarded, the remaining eluate being collected. The cartridge was then washed with methanoljdich loromethane (5:95, 15 ml), the eluate being added to that already collected. The combined eluates were subsequently evaporated to dryness under vacuum, the residue redissolved in chloroform/hexane (1:1, 2 ml) and transferred to a Bond Elut cyano extraction column. The toxins were eluted from the column with chloroform/hexane (1:1, 13 ml) and the eluate evaporated to dryness under vacuum. The residue represented a 10 g equivalent sample weight. A 2 g equivalent sample weight of the residue was placed in a second vial and retained for chromatographic analysis.
2.3.2.2 Sample preparation - (Sydenham and Thiel, 1987)
This method was initially developed for the analysis of fun gal cultures, but was also found to be suitable for the de..,. 33
termination of diacetoxyscirpenol and T-2 toxin in grain
samples. ·
A subsample of ground maize (25 g) was extracted with metha nol/water (1:1, 100 ml), by agitating at high speed for 30 minutes on a wrist action shaker. An aliquot (20 ml) of the filtered extract was transferred to an Extrelut 20 column and left for several minutes for adsorption purposes. The toxins were eluted with cyclohexanejdichloromethane (1:3, 40 ml), the eluate being collected and evaporated to dryness. The dried residue, dissolved in chloroform/hexane (1:3, 8 ml) was transferred to a silica gel Sep-pak cartridge. The cartridge was washed with benzene (6 ml) and acetonejbenzene (5:95, 8 ml). The purified toxins were eluted from the cartridge with diethyl ether (6 ml), the eluate being collected and evapora- ted to dryness under vacuum. This residue represented a 5 g equivalent sample weight.
2.3.3 Analytical standards diacetoxyscirpenol and T-2 toxin
The second method evaluated (Sydenham and Thiel, 1987) incor porated the use of another type A trichothecene (neosolaniol monoacetate - Table 1.1) as an internal standard prior to chromatographic separation. Diacetoxyscirpenol, T-2 toxin and neosolaniol monoacetate stock solutions were prepared in -1 glass-distilled methanol (500 ~g ml ). Aliquots (20 ~1) of the diacetoxyscirpenol and T-2 toxin stock solutions were placed in one vial while a similar aliquot of the neosolaniol 34
monoacetate stock solution was placed in a separate vial.
The contents of each vial were evaporated to dryness and derivatized as outlined in section 2.3.4. Following derivati- zation, 5, 10, 15 and 20 tJ.l aliquots of the organic phase containing the standard toxin esters were placed in each of four separate vials to which 15 tJ.l of the organic phase containing the ester of the internal standard were added. The contents were evaporated to dryness under nitrogen, redissolved in hexane (1 ml)t capped, mixed and analysed by capillary gas chromatography. A calibration curve (covering the range 50 - 200 ng ml -1 diacetoxyscirpenol and T-2 toxin with 150 ng ml-1 internal standard) was constructed for each toxin by plotting the ratio of the toxin peak areas to that of the internal standard against toxin concentration.
2.3.4 Derivatization procedure - type A trichothecenes
Both methods required the formation of the heptafluorobutyryl derivatives of the purified extracts prior to capillary GC separation
The residues (transferred to 4 ml vials) were dissolved in toluene/acetonitrile (95;5, 1 ml) and heptafluorobutyrylimi dazole (HFBI, 100 tJ.l) was added. The vials were capped, the contents mixed and heated at 60°C for 1 hour. After cooling, sodium phosphate buffer (0.1M, pH 6.0, 1 ml) was added (to neutralize any excess HFBI) and the contents •ixed. ~he 35
phases were allowed to separate and a suitable aliquot of the
upper ·organic phase was removed, evaporated to dryness and redissolved in hexane (1 ml), prior to analysis by capillary gas chromatography. All samples were chromatographed on the same day as derivatization.
2.3.5 Chromatographic separation - type A trichothecenes
Samples were chromatographed using a Carlo Erba Model 5300 high resolution gas chromatograph, equipped with a 63Ni elec tron capture detector (ECD) used in the constant frequency mode, split injector and a 25m x 0.32 mm (i.d.) fused silica capillary column coated with a 0.25 ~m film thickness of PS 255. Data were collected using a Waters 745 data module. The exact chromatographic conditions are given in Table 2.1.
2.4 ANALYTICAL METHODOLOGY - TYPE B TRICHOTHECENES
Two methods for the simultaneous determination of nivalenol and deoxynivalenol in maize were e~aluated {Tanaka et al., 1985a; Scott et al., 1986).
2.4.1.1 Sample preparation- (Tanaka et al., 1985a)
A subsample (20 g) was extracted with acetonitrile/water (3:1, 200 ml) by shaking for 30 minutes on a wrist action shaker. The mixture was then filtered and an aliquot of the filtrate {125 ml) was partitioned with hexane (100 ml). The 36
TABLE 2.1 CHROMATOGRAPHIC PARAMETERS FOR THE DETERMINATION
OF DIACETOXYSCIRPENOL AND T-2 TOXIN
Column PS-255 25m x 0.32 mm (i.d.) Film thickness 0.25 J,.Lm Carrier gas (flow rate) Helium at 35 em sec-1 Make-up gas (flow rate) Nitrogen at 30 ml min-1 Detector ECD ( 63Ni) Attenuation 64 Injection volume 1 J,.Ll Split ratio 1 : 10 Injector temperature 200°C Detector temperature 300°C Temperature profile (1) 180°C - 220°c at 4°C min-1 (2) 220°c - 250°C at 15°C min-1 (3) 250°C for 5 min
upper hexane layer was discarded and ethanol (120 ml) was added to the aqueous acetonitrile layer and evaporated to dryness under vacuum at 45 0 c. The residue was dissolved in methanol (5 ml) and a 4 ml aliquot was transferred to the top of a wide bore chromatography column (30 em x 2.2 em i.d.), containing Florisil (20 g) packed between two layers of anhydrous sodium sulphate (2 x 10 g), in toluene. The column was washed with hexane (200 ml) and the toxins eluted with chloroform/methanol (9:1, 250 ml). The eluate was evaporated to dryness and redissolved in methanol (2 ml). An aliquot of the extract (1 ml) was mixed with chloroform (1 ml) and 37
hexane {50 ml), and placed on a silica-gel Sep-Pak cartridge.
The cartridge was washed with n-hexane (50 ml) and the toxins eluted with ethyl acetate;toluenejmethanol (20:5:1, 100 ml). The eluate was collected, evaporated to dryness, redissolved
L in methanol (2 ml) and an aliquot (1 ml) was reserved for derivatization and separation by capillary GC.
2.4.1.2 Sample preparation- (Scott et al., 1986)
A milled subsample (50 g) was extracted with methanol/water (7:3, 150 ml) by blending at high speed for 5 minutes in a Sorvall omnimixer. The "puree" was transferred to a centri- fuge tube and centrifuged for 15 minutes at 2000 rpm. An ali quot of the clear supernatant (30 ml) was mixed for 2 minutes with 10% aqueous ammonium sulphate (120 ml) and diatomaceous earth, using a magnetic stirrer. The contents were then fil- tered and an aliquot of the filtrate (10 ml) transferred to an Extrelut 20 column. Following a period of time for absorption (± 5 minutes), the toxins were eluted with ethyl acetate (16D ml), the eluate being collected and evaporated to drynes~ under reduced pressure. The residue was dissolved in dichloromethanejmethanol (3:1, 0.5 ml) and placed on a small diameter chromatography column (10 mm i.d.) containing silica gel (2 g) embedded between two layers of anhydrous sodium sulphate (2 x 1 g), in toluene. The column was washed / successively with toluene (15 ml) and hexane(15 ml), before the purified toxins were elu~ed with dichloromethanejmethanol (9:1, 60 ml). The latter eluate was collected, evaporated to 38
dryness under reduced pressure, transferred to a 4 ml capaci-
ty vial and dissolved in dichloromethanejmethanol (3:l, 1.33 ml). The solution represented a 0.5 g ml-1 sample equivalent
weight. An aliquot of the solution (0.5 ml) was removed to a
second vial and evaporated to dryness pri,or to derivatization and separation by capillary GC.
2.4.2 Analytical standards - nivalenol and deoxynivalenol
-1 Nivalenol and deoxynivalenol stock solutions (50 ~g ml ) . were prepared in glass distilled methanol. Aliquots of the
stock solutions (50 ~1) ·Were placed in a vial and evaporated to dryness prior to derivatization (section 2.4.3). Following
derivatization, 2.5, 5, 10 and 15 ~1 aliquots were placed in
separate vials and diluted to 500 ~1 with glass distilled . . -1 hexane to give a concentration range of 25 - 150 ng ml per
toxin. A calibration curve for each toxin was constructed by plotting the peak area of each toxin against toxin concentra tion.
2.4.3 Derivatization procedure -·nivalenol and deoxynivale- nol
The type B trichothecenes also require derivatization due to their chemical characteristics.
Tri-Sil TBT (a commercially available mixture of N-trime-
thylsilylimidazole: N,O-bistrimethylsilylacetamide: Trime- 39
thylchlorosilane, 3:3:2, v;v, 50 ~1) wris added to each vial
which was then capped, mixed and allowed to stand for 15
minutes at room temperature. Hexane (500 ~1) was then added
to each vial followed by phosphate buffer (0.1M, pH 7, 2 ml).
The contents of each vial ~e~e mixed, the layers allowed to separate, and a suitable aliquot of the upper organic phase removed to a second vial where the volume was made up to 500
~1 with hexane.
~ABLE 2.2 CHROMATOGRAPHIC PARAMETERS FOR THE DETERMINATION OF NIVALENOL AND DEOXYNIVALENOL
Column DB-5 30m x 0.32 mm (i.d.)
Film thickness 0.25 ~m 1 . Carrier gas (flow rate) Helium at 35 em sec- Make-up gas (flow rate) Nitrogen at 48 ml min~ 1 63 Detector ECD ( Ni) Attenuation 32
Injection volume 1 ~1 Splitless (on time) 0.5 min Injector temperature 180°C Detector temperature 300°C Temperature profile (1) ao 0 c for 1 min (2) ao 0 c - 275°C at 10°c min-1 40
2.4.4 Chromatographic separation - nivalenol and deoxyniva
lenol
All samples were chromatographed using the instrumentation outlined in section 2.3.5. However, the capillary column used was a 30 m x 0.32 mm i.d fused silica DB-5 (0.25 IJ,m thick ness). The exact chromatographic conditions are given in Table 2.2.
2. 5 ANALYTICAL METHODOLOGY - ZEARALENONE
Three methods for the determination of zearalenone in maize were evaluated (Association of Official Analytical Chemists {AOAC), 1984; Chang and DeVries, 1984; Bagneris et al., 1986). The methods differed considerably with respect to their extraction and clean-?p methodologies.
2.5.1.1 Sample preparation - (AOAC, 1984)
A subsample (50 g) was extracted with chloroform/water (10:1, 270 ml), to which diatomaceous earth (25 g) had been added, by shaking for 30 minutes on a wrist action shaker. The mix ture was filtered and an aliquot of the filtrate (50 ml) was placed onto a chromatography column packed with silica gel (10 g) placed between layers of anhydrous sodium sulphate (2 x 5 g), in chloroform. the column was washed successively with hexane (150 ml) and benzene (150 ml) before the toxin was eluted with acetone/benzene (5:95, 250 ml). The latter 41
eluate was collected and evaporated to dryness under reduced
pressure. The residue, redissolved in hexane (40 ml) was transferred to a small separating funnel. Zearalenone was partitioned into acetonitrile (3 x 10 ml), this layer being collected, evaporated to dryness and transferred to a vial in benzene (500 ~1).
2.5.1.2 Sample preparation - (Chang and DeVries, 1984)
A subsample (50 g) was extracted with dichloromethanejwater (10:1, 270 ml) to which cupric carbonate (10 g) was added, by shaking for 30 minutes. The mixture was filtered and an aliquot of the filtrate (50 ml) was evaporated to dryness under reduced pressure. The residue was r~dissolved in aceto nitrile (80 ml) and partitioned with petroleum ether (100 ml). The acetonitrile layer was collected and evaporated to dryness under reduced pressure. The residue was subsequently redissolved in methanoljwater (4:1, 2 ml), prior to separtion by HPLC.
2.5.1.3 Sample preparation- (Bagneris et al., 1986)
A subsample (30 g) was extracted with chloroformjwater (12.5:
1, 270 ml) by shaking for 15 minutes. The mixture was then centrifuged for 3 minutes at 2000 rpm. The extract was fil tered and an aliquot of the filtrate (50 ml) transferred to a separatory funnel. Saturated sodium chloride solution {10 ml) and aqueous sodium hydroxide solution (2%, 50 ml) were then 42
added to the funnel, the contents shaken and the two layers allowed to separate. The lower organic layer was discarded and the aqueous layer partiti.oned with chloroform (50 ml) •
The organic layer was once more discarded and citric acid solution (0.5M, 50 ml) added to the funnel. Zearalenone was extracted from the solution by partitioning with dichlo romethane (2 x 50 ml). The dichloromethane extracts were dried with anhydrous sodium sulphate, collected and evapora ted to dryness under.reduced pressure. The residue was redis solved in methanol/water ( 7: 3 ,· 500 JJ.l) , prior to separation by HPLC.
2.5.2 Analytical standard - zearalenone
-1 Zearalenone stock solution (50 J,Lg ml ) was prepared in methanol/water (7:3). Aliquots (0.25, 0.5, 0.75 and 1 ml) were diluted to 10 ml in separate volumetric flasks toyield -1 a concentration range of 1.25 - 5 J,Lg ml • These concentra- tions were verified by recording their UV absorptions at 236 nm and calculating the concentration using the molecular weight and molar extin~tion coefficients previously published - . (Cole and Cox, 1981) and the formula:
Zearalenone = Abs (@ 236 nm) x 1000 x molecular wt. (318) concentration 1 (JJ.g ml- ) € (@ 236 nm) = 29700 43
2.5.3 Chromatographic separation - zearalenone
Reverse phase separations of 20 ~1 aliquots of standards and samples were performed on a 4-~m Nova-Pak c18 column (15 em x 3.9 mm i.d.). Mobile phase was delivered by means of a.Waters Model 510 liquid chromatographic pump. Peak detection was performed on a Perkin-Elmer 650S fluorescence detector. The prevailing chromatographic conditioris are presented in Table 2.3. Data were collected on a Waters 745 data module and quantitative determinations were by comparison of toxin peak areas against a calibration curve.
TABLE 2.3 CHROMATOGRAPHIC PARAMETERS FOR THE DETERMINATION OF ZEARALENONE
Column Nova-Pak RP c18 (4-~m) 15 em x 3.9 mm (i.d.) Mobile phase Methanol : water (7:3) . -1 Flow rate 0.5 ml m1n
Injection volume .20 ~1 Detector Fluorescence Detector settings 274 nm (Excitation) 465 nm (Emission) Slit widths 15 nm 44
2.6 ANALYTICAL METHODOLOGY - MONILIFORMIN
Two methods for the determination of moniliformin in maize were evaluated (Thiel et al., 1982b; Scott and Lawrence,
1987) 0
2.6.1.1 Sample preparation- (Thiel et al., 1982b)
Sample extracts were prepared by extracting a subsample (3 g) with distilled water (40 ml) in a centrifuge tube for 1 hour.
The contents were then centrifuged at 10,000 g, and the super natant extracts filtered through Millipore HA filters (0.45 J..Lm). The filtered extracts were retained for HPLC separation.
2.6.1.2 Sample preparation - (Scott and Lawrence, 1987)
Extracts were prepared by extracting a subsample (10 g) with acetonitrile/water (95:5, 100 ml) by blending at high speed for 5 minutes. The samples were filtered and partitioned with hexane (150 ml). The hexane layer was discarded and an aliquot (50 ml) of the remaining phase evaporated to dryness under reduced pressure. The residue was redissolved in metha nol (2 ml). An aliquot (250 J..Ll) of the partially purified ex tract was placed on a pre-conditioned Sep-pak c18 cartridge and moniliformin was eluted with distilled water (2 ml). The eluate was collected, evaporated to dryness and the residue dissolved in 500 IJ.l of the mobile phase {Table 2.4). The so lution was passed through a small prepacked alumina/Millipore 45
HA filter column by syringe pressure, the eluate being col lected and retained for HPLC analysis.
2.6.2 Analytical standard - moniliformin
Moniliformin stock solution (50 ~g ml-l) was prepared in dis tilled water. Aliquots (0.5, 1.0, 1.5, and 2.0 ml) of .the stock solution were placed in separate 10 ml volumetric flasks and diluted to the mark with distilled water. The con centrations were verified by measuring their absorbances at 229 nm and applying the formula given in section 2.5.2, sub stituting the values of molecular weight and molar extinction coefficient for moniliformin (Cole and Cox, 1981).
TABLE 2.4 CHROMATOGRAPHIC PARAMETERS FOR THE DETERMINATION . OF MONILIFORMIN (PAIRED ION CHROMATOGRAPHY)
Column Nova-Pak RP c18 (4-~m) 15 em x 3.9 mm (i.d.) Mobile phase 0.005 M Tetrabutylammonium hydrogen sul phate in 4% methanol/0.1 M phosphate buffer (pH 7.0) Flow rate 1 ml.min -1
Injection volume 20 ~1 Detector uv Detector setting 229 nm Attenuation 0.02 AUFS 46
2. 6. 3 · Chromatographic separation - moniliformin
Purified extracts were separated on a 4-~m Nova-Pak c18 co lumn (15 em x 3.9 mm i.d.). Mobile phase was delivered by means of a Waters Model 510 liquid chromatographic pump and peak detection was performed on a Waters Model 481 variable wavelength u.v. detector. Data were collected with a Waters 745 data module. The exact chromatographic conditions are given in Table 2.4. Quantitative measurements of moniliformin were obtained by the comparison of toxin peak areas in each sample to a calibration curve of standard peak areas covering -1 the range 2.5 - 10 ~g ml •
. 2 • 7 INNOVATIVE METHODOLOGY
2.7.1 Sample preparation- fumonisin B1
Sample extracts were prepared by extracting a subsample (25 g) ··of maize (5 g for culture material) with methanol/water (3:1, 50 ml) by blending at high speed for 5 minutes and fil tering. An aliquot of the filtrate (25 ml) was evaporated to dryness at 50°C, redissolved in methanoljwater (1:3, 25 ml) and partitioned with chloroform (2 x 50 ml). The organic phase was discarded and the aqueous phase evaporated to dry ness and redissolved in methanol/water (1:3, 10 ml; 50 ml for cultures). Aliquots of these extracts (2 ml for HPLC and 1 ml for GC) were further purified by application to a Waters Sep
pak c18 cartridge, washing with methanoljwater (1:3, 10 ml) 47
and eluting the toxin with methanoljwater (3:1, 10 ml). The eluate was evaporated to dryness and retained for derivatiza tion prior to chromatographic separation.
2.7.2 Derivatization for HPLC separation
Mal~yl derivatives of the purified extracts were prepared and analysed by HPLC according to a previously published method (Siler and Gilchrist, 1982), with minor modifications. The residues were dissolved in sodium carbonate buffer solution (0.05M, pH 9,2, 5 ml) and treated with an excess of 1M maleic anhydride solution in dioxane (3 x 10 ~1). The pH of the solution was constantly monitored while the solution was agitated by means of a magnetic stirrer. The pH was adjusted to >9 using sodium hydroxide solution (O.lM), following each addition of maleic anhydride. Following a reaction time of 10 minutes, the pH of the solution was adjusted to between 6 - 7 using dilute HCl (O.lM). The derivatized extracts were then evaporated to dryness under reduced pressure at 50°C, redis solved in methanoljwater (1:1, 2 ml) and retained for HPLC separation.
2.7.3 Hydrolysis, derivatization and GC separation
Purified sample extracts were hydrolysed (under nitrogen) with HCl (6M, 2 ml) at 95°C for 3 hours. Each sample was then cooled, centrifuged at 2000 rpm and an aliquot of each sub jected to a previously published esterification procedure 48
(Labadarios et al., 1984). Briefly an aliquot (500 ~1) of
each hydrolysate was transferred to an appropriate tube and the excess acid removed under reduced pressure. Esterification of the residue was performed under nitrogen
using isobutanol containing 3M HCl (250 ~1) and heating at 100°e for 45 minutes. The acidified isobutanol was removed under reduced pressure and the residue acylated with
heptafluorobutyric anhydride (100 ~1) by heating for 10 minutes (under nitrogen) at 150°C. The samples were then cooled in ice, freeze-dried, dissolved in ethyl acetate (100
~1) and transferred to small vials prior to separation by capillary GC.
2.7.4. Secondary extraction and derivatization- fumonisin B1
A secondary technique for the determination of fumonisin B1 was investigated which involved extraction, fluorescamine derivatization and HPLC separation.
Samples of maize (25 g) were extracted with methanol/water (3:1; 50 ml) by shaking for 30 minutes. The extracts were filtered and an aliquot of the filtrate (5 ml) was evaporated to dryness at 50°C. The residue was partially redissolved in methanol/water (7:3, 1 ml) and fully solvated following the addition of methanol (1 ml). The solution was transferred to a prepared chromatographic column (10 mm i.d.) containing activated silica gel (2 g; 70-230 mesh) suspended between two layers of anhydrous sodium sulphate (2 x 1 g), in methanol. 49
The column was washed with methanol (15 ml) the toxins eluted
with 0.1% acetic acid/methanol (20 ml) and the eluate was
evaporated to dryness~ The purified residues were redissolved in· 0.05M sodium hydrogen carbonate (pH 8.6; 1 ml) and an .aliquot (25 IJ.l) was diluted to 500 IJ.l in a plastic microfuge tube (1.8 ml capacity) with the same solution. While vortex -1 mixing, fluorescamine (50 IJ.l; 2 mg 100 j.J.l in acetone; Aldrich) was added to the tube. Mixing was continued for 1 minute and the mixture was then centrifuged for 1 minute at
9,000 g. The derivatized extracts were retained for HPLC separation coupled with fluorescence detection.
Analytical standard - fumonisin B 2.7.5 1
A stock solution of fumonisin B1 was prepared in distilled -1 water (2.5 mg ml ). An aliquot of the stock solution (200
IJ.l) was used to prepare the standard maleyl derivative as outlined in section 2.7.2, to give a final concentration of -1 . 250 IJ.g ml . A second aliquot of the same stock solution was used to prepare the fluorescamine derivative (as outlined in 1 section 2.7.4), to give a final concentration of 3.25 IJ.g ml- .
Tricarballylic acid (TCA) standard stock solution was prepa -1 red in glass distilled methanol (10 mg ml ). An aliquot of this solution (100 IJ.l) was used to prepare a standard esteri- fied derivative as outlined in section 2.7.3. Aliquots.of the derivative in ethyl acetate (2.5, 5, 7.5 and 10 IJ.l) were transferred to separate vials and diluted with ethyl acetate 50
-1 (1 ml) to give a concentration range of 25 - 100 ~g ml TCA
as its butyl ester.
2.7.6 Chromatographic conditions - fumonisin B (as its ma 1 leyl derivative) by HPLC
HPLC separations of the preformed maleyl derivatives were performed on a 4-~m Nova-pak c18 column. Solvent delivery and peak detection was performed using the equipment outlined in section 2.6.3. The prevailing chromatographic conditions used are given in Table 2.5. Quantitative determination of
fumonisin B1 was by comparison of peak areas in the samples to that of the peak area of a similar derivatized fumonisin
B1 standard.
TABLE· 2.5 I CHROMATOGRAPHIC. PARAMETERS FOR THE DETERMINATION
OF FUMONISIN B1 (AS ITS MALEYL DERIVATIVE) BY HPLC
Column Radial-Pak RP c18 (4-~m) 10 em x 8 mm (i.d.) Mobile phase Methanol 0.05M phosphate buffer (7:3) adjusted to pH 3.5 with concentrated HCl Flow rate 1.5 ml/minute
Injection volume 5 or 10 ~l Detector uv Detector setting 230 nm Attenuation 0.02 AUFS 51
.2.7.7 Chromatographic separation - fumonisin a1 (as its fluorescamine derivative) by HPLC
Aliquots of the preformed fluorescamine derivatives were se parated on a 7 ~m Phenomenex Ultracarb ODS 30 c column. Sol vent delivery was performed using the equipment outlined in section 2.6.3 and peak detection was performed using a Perkin
Elmer 650S fluorescence detector. The prevailing chromato graphic conditions used are detailed in Table 2.6.
TABLE 2.6 CHROMATOGRAPHIC PARAMETERS FOR THE DETERMINATION
OF FUMONISIN B1 (AS ITS FLUORESCAMINE DERIVATIVE)
Column Phenomenex Ultracarb ODS 30 (7 ~m) 25 em x 4.6 mm (i.d.) Mobile phase O.lM Acetate buffer (pH 4.0) Acetoni-
trile (1:1)
Flow rate 1 ml/min
Injection volume 20 ~1 Detector Fluorescence Detector settings 390 nm (Excitation) 475 nm (Emission) Slit widths 20 nm 52
.2.7.8 Chromatographic conditions -.TCA by capillary GC
The butyl esters of standards and samples were separated on a
DB-5 capillary column. The Carlo-Erba Model 5300 gas chroma
tograph used was equipped with a splitless injector and a
flame ionization detector (FID). The exact chromatographic
conditions are given in Table 2.7. Data were collected with a Waters 745 data module, and quantitative determinations of butyl esters of TCA present in the samples, 'was by comparison
of peak areas to that of a calibration curve of the peak areas of similarly derivatized standards.
TABLE 2.7 CHROMATOGRAPHIC PARAMETERS FOR THE DETERMINATION OF TRICARBALLYLIC ACID
Column DB-5 30 m x 0.32 mm (i.d._)
Film thickness 0.25 !J.m -1 Carrier gas (flow rate) Helium at 35 em sec Detector FID
Attenuation 8
Injection volume Splitless (on time) .0.5 min Injector temperature 140°C
Detector temperature 280°C 1 Detector fuel gases Air at 300 ml min- " 1 Hydrogen at 30 ml min-
Temperature profile (1) 50°C for 3 min 1 (2) .50°C - 280°C at 15°C min- 53
2.8 CONFIRMATORY TECHNIQUES
2.8.1 Improvement to chromatographic data
In order to improve or Verify some of the chromatographic ob servations concerning the presence of the selected Fusarium mycotoxins in the series of Transkeian maize samples, various alterations and/or additional confirmatory techniques were used.
Possibly the easiest of these techniques (spiking) was used in each analysis. Briefly, mixtures were prepared containing aliquots of the sample extracts together with aliquots of si milarly prepared (ie. derivatized where necessary) authentic target toxin standards. Aliquots of these mixtures were then re-chromatographed, and the resultant chromatogram was com pared to that obtained for the sample extract. Other techni ques andjor improvements used in this investigation are pre sented in sections 2.8.2 - 2.8.6.
2.8.2 Chromatographic confirmation of nivalenol and deoxyni
valenol
Neither method evaluated (Tanaka et al., 1985a; scott et al., 1986) made use of an internal standard. A type B tricho thecene, fusarenon-X, whose characteristics and structure are similar to both nivalenol and deoxynivalenol (Table 1.2) was incorporated as an internal standard for this study. A 54
Tri-Sil TBT derivative of fusarenon-X was prepared as pre- viously described (section 2.4.3) and an aliquot (correspon 1 ding to a final concentration of 600 ~g rnl- ) was added to each sample vial prior to chromatographic separation.
The presence of nivalenol and deoxynivalenol in several sam- ples extracts was confirmed by capillary GC/rnass spectrometry
(GC/MS) using a Finnigan Model 4500 GC/MS and the conditions outlined in Table 2.8.
TABLE 2.8 CHROMATOGRAPHIC PARAMETERS FOR THE CONFIRMATION OF TRICARBALLYLIC ACID, NIVALENOL AND DEOXYNIVALENOL
Column DB-5 60 rn x 0.32 rnrn (i.d.)
Film thickness 0.25 ~rn -1 Carrier gas (flow rate) Helium at 30 ern sec oetector Finnigan 4500 MS
Injection volume 1 ~1 Splitless (on time) 0.5 min Injector temperature 180°C Detector temperature 300°C Temperature profile (1) 80°C for 1 min 1 (2) 80°C - 280°C at 15°C rnin-
Mass range 100 - 400 rnjz 55
2.8.3 Chromatographic confirmation of zearalenone
Extracts of a number of samples that indicated relatively high levels of zearalenone were re-chromatographed (using the chromatographic conditions detailed in Table 2.3) and moni tored at three different UV wavelengths simultaneously, using a Hewlett-Packard 1040A diode array detector. The wave lengths monitored corresponded to the three wavelength maxi ma recorded for the UV spectrum of a zearalenone standard
(Figure 3.3). The measured ratios of the wavelengths of th~ zearalenone peak in the sample extract were then compared with the same ratios measured in the standard. In addition, the extracts were also monitored over the wavelength range 200-400 · nm and computer software accompanying the instrument was used to display the data as a 3 dimensional representa tion.
2.8.4 Chromatographic confirmation of moniliformin
In addition to the paired ion chromatographic method for the determination of moniliformin in sample extracts (Table 2.4), several of the extracts were also chromatographed using ion exchange chromatography (the prevailing chromatographic con ditions are given in Table 2.9). Sample extracts, separated by both techniques were also monitored using the diode array detector as previously described (section 2.8.3). The wave length maxima monitored were conspecific with those deter mined from the UV spectrum of authentic moniliformin (Figure
3. 2) • 56
TABLE 2.9 CHROMATOGRAPHIC PARAMETERS FOR THE CONFIRMATION OF
MONILIFORMIN BY ION EXCHANGE CHROMATOGRAPHY
Column Whatman RAC II SAX (5 J.Lm) 10 em X 4.6 mm Mobile phase 0.001M Sodium dihydrogen phosphate
Flow rate 1.5 ml min-1
Injection Volume 20 JJ.l Detector uv Detector setting 229 nm Attenuation 0.02 AUFS
TABLE 2.10 CHROMATOGRAPHIC PARAMETERS FOR THE CONFIRMATION
OF FUMONISIN B1 (AS ITS MALEYL DERIVATIVE) BY HPLC
Column Phenomenex Ultracarb ODS 30 (7 J.Lm) 25 em x 4.6 mm (i.d.) Mobile phase O.OSM Acetate buffer (pH 3.5) : methanol (35:65) pH maintained at 3.5 with concen- trated HCl 1 Flow rate 1 ml min-
Injection volume 20 IJ.l Detector uv Detector setting 230 nm Attenuation 0.02 AUFS 57
2.8.5 Chromatographic confirmation of fumonisin B 1
( Extracts prepared by the maleyl derivative procedure were separated on a second reverse phase column using an acetate
buffer mobile phase system (details are given in Table 2.10).
2.8.6 Chromatographic confirmation of TCA
Butyl esters of several sample extracts, prepared as pre
viously described (section 2.7.3) were subjected to GC/MS analysis. Confirmation of the presence of TCA as its butyl
derivative in purified sample extracts was by comparison of their MS to that .of the. MS of a similarly prepared TCA
analytical standard. The GC/MS used was the same as that described in section 2.8.2., and the prevailing chromatograp hic conditions were the same as those for the confirmation of
the presence of nivaienol and deoxynivalenol (Table 2.8).
2.9 STATISTICAL ANALYSES
Statistical analyses of the data were performed by Dr D.J. van Schalkwyk via the analysis of variance, followed by the student-Neuman-Keuls multiple comparison method. Since a number of the variables included either percentages or. the means of percentages~ it was expected that they might not be normally distributed. As a precaution, both parametric and non-parametric analyses were performed. Pearson parametric analyses of the data were done by .the analysis of variance 58 and, where necessary, followed by the Student-Neuman-Keuls multiple comparison method. In addition, Spearman non-parame tric analyses were also performed. The results were consis tent and the probability levels'reported are the most conser vative (ie least significant) obtained. 59
CHAPTER 3
RESULTS AND DISCUSSION
A: STANDARDS AND EXISTING METHODOLOGY
3.1 ANALYTICAL STANDARDS
3.1.1 Importance of authentic standards
Several factors have to be taken into consideration to esta blish a precise and accurate analytical determination, such as the degree to which the subsample used for the analysis is representative of the bulk of the sample. The availability of an "authentic" analyte standard is, however, an absolute pre requisite, especially in quantitative chromatographic techni ques. This may be illustrated by the fact that in the absence of sophisticated monitoring techniques (i.e. MS) at the final chromatographic stage of an analysis, the presence or absence and concentration of a particular analyte, in an often com plex matrix, is solely based on the retention time and size of a chromatographic peak, and its comparison to the retention time and size of a similarly chromatographed "authentic" standard.
Some of the standards used in this study were obtained from commercial sources, but several (including deoxynivalenol, 60
neosolaniol monoacetate, moniliformin and fumonisin B ) were 1 previously isolated within the Institute. In order to verify the identity and purity of the analytical standards used, se- veral analytical parameters were measured on each toxin (de-
tailed in sections 3.1.2 3.1.8). Where possible the re- sults were compared with those previously published (Cole and Cox, 1981; Pohland et al., 1982). It was not possible to collect all data on each toxin due to insufficient quantities of a number of toxin standards.
3.1.2 Melting points
Where possible, approximately 20 mg of each toxin was dried
0 in an oven at 110 c for 1 hour, r~moved and stored in a des- sicator. Several crystals of each toxin standard were placed between two cover slides and their melting points determined using a Reichert melting point variable heating stage, fitted with a single lens microscope. A large glass plate was placed over the heating stage, following the placement of each sample~ so as to ensure an enclosed environment within the heating block. The temperature of the stage was increased gradually and the temperatures at which the crystals of each ·toxin appeared to start melting and were completely molten, were recorded. The melting points were performed in tripli- cate and the melting point ranges recorded are given in Table 3.1. The accuracy of the equipment used to determine the mel- ting points was checked against the melting point of benzoic acid. 61
TABLE 3.1 MELTING POINTS OF SELECTED MYCOTOXIN STANDARDS
Toxin Melting Point (°C) Published Melting
Point (°C)
1 Deoxynivalenol 152 - 155 151 - 153 1 Diacetoxyscirpenol 163 - 165 162 - 164 2 Neosolaniol mono 188 - 189 190 - 190.5 acetate 1 T-2 toxin 146 - 149 151 - 152 2 145 - 149 1 Zearalenone 163 - 165 164 - 165 2 161 - 164
1 Cole and Cox, (1981) 2 Pohland et al., (1982)
3.1.3 Thin layer and high perfo~ance liquid chromatography
Toxin standard solutions were screened by both TLC and HPLC techniques on either silica gel TLC plates or reverse phase
HPLC columns, as previously described (Tables 2.~ and 2.4). A number of solvent systems were used for TLC separations and the chromatographed plates were sprayed (if necessary) with 62 p-anisaldehyde (0.5% v;v in methanol:acetic acid:sulphuric acid, 85:10:5) solution as well as with sulphuric acid solu- tion. Where spraying of the plates was necessary, the plates were heated in an oven at 110°C prior to visua1 inspection. Plates which did not require spraying were viewed under either longwave (366 nm) or shortwave (254 nm) UV light. In each case, only a single spot (TLC) or peak (HPLC) could be detec ted.
3.1.4 capillary GC analyses of HFBI and Tri-Sil TBT deriva
tlves
HFBI and Tri-Sil TBT derivatives of each trichothecene stan- dard used in this study were prepared as described in sec tions 2.3.4. and 2.4.3., respectively, at concentrations of approximately 100 ~g ml-1 . Splitless injections of each deri vative were separated on a Carlo-Erba GC fitted with an apo lar Machery-Nagel fused silica capillary column (25 m x 0.32 mm i.d.) coated with a 0.25 ~m thickness of SE-30. The pre- vailing chromatographic conditions used were the same as those previously described in Table 2.2. Using the FID as a universal detector, only a single chromatographic peak could be detected in each of the derivatives.
3.1.5 Infra-red absorption spectra
Infra-red spectra of a number of toxin standards were collec ted using a Perkin-Elmer 983 IR spectrophotometer. The reso- 63
-1 lution was 3 em , and each sample was scanned over the range 1 4000 - 180 cm- Each toxin standard was thoroughly ground with spectroscopic grade potassium bromide (KBr) at a mass -1 fraction of approximately 1.25 mg g KBr. The ground mixtures were pressed into 13 mm discs prior to analysis. The IR spectra obtained were compared to those previously published (Cole and Cox, 1981, Pohland et al., 1982). Excellent agreement was obtained for each sample analysed which included deoxynivalenol, nivalenol, T-2 toxin, diacetoxyscirpenol, neosolaniol monoacetate, moniliformin and zearalenone. Figure 3.1 shows the comparison of the spectrum obtained for the diacetoxyscirpenol standard to that reported in the li terat.ure ( Pohland et al., 1982).
3.1.6 Ultra-violet absorption spectra
Standard solutions of toxins were prepared in methanol, water or a mixture of the two solvents. The spectra were recorded on a Beckman 5260 double beam recording spectrophotometer 1 over the range 200-350 nm at a scan speed of 2 nm sec- and a -1 chart speed of 20 nm inch . Spectra were once again compared with those reported in the literature (Cole and cox, 1981;
Pohland et al., 1982). In each case excellent agreement was obtained between the generated spectrum and that published.
Figures 3.2 and 3.3 show the spectra obtained for monilifor min and zearalerione, respe6tively. The concentrations of the toxins in their respective solutions are given in the legends to each figure. -.....c.-~ .. • , •• IOQU.»IO• ··:·i-::··1---: ·I I···-- ··-1-··1 -1------·-\ , ___;A·+· . I . I . I "' '\. • . . . /''1'111'~IV" h-.-; I ' .
=.. ·· · h--1···11---1-··· I·· ·: ···). 1"\· nl,- 11f· ·I·· --'~, ..H --
1~ @l~S~fi 1;~~~~~f11~:fj CICIO UIIO XIIO 1D xatJ-IiooiGil'eiOIIO ____ I I __...."""''
~------~--~----~------~------~------~------~--~100
i= 80 zw u 01!w ~ 60 w uz
~ 40 ~
~ 20
0 4000 3000 2000 1600 1200 800 400 200
WAVENUMBER(CM ')
FIG. 3.1 Infra-red absorption spectrum of diacetoxyscirpenol with - insert 0'1 ~ published spectrum {Pohland et al., 1982) 65
1.8
1.6 229nm
1.4
1.2
~ Q 1.0 :ae- ~ 0.8
0.6 254nm 0.4
0.2
0
190 230 270 310 350
Wavelength (nm)
1 FIG. 3.2 Ultra-violet spectrum of moniliformin [9.25 IJ.g ml- ] in water
3.1.7 Fluorescence spectroscopy
Fluorescence characteristics were measured using a Perkin- Elmer 650S fltiorescence spectrophotometer. The excitation
~pectra (200 - 450 nm) were recorde~ by setting the emission spectral bandpass (slit width) to 15 nm and the emission dial to zero order. The excitation bandpass was set at approxi- 66
1.8 236nm 1.6
1.4
1.2
r:l 0 1.0 ~e- ~ 0.8 274nm
0.6
0.4 316nm
0.2
0
190 230 270 310 350
Wavelength (nm)
-1 FIG. 3. 3 Ultra-violet spectrum of zearalenone [17 ~g ml ] in methanol:water (7:3)
mately 3 nm and spectra were recorded at a scan rate of 60 nm minute-1 . Emission spectra (350-550 nm) were recorded at the same scan rate, by setting the excitation wavelength to that
corresponding to the excitation maxima and the.spectral band . passes were reversed. Of the available standards, only zeara lenone displayed any fluorescence characteristics which are illustrated in Figure 3.4. a b
129% 129'<
I I 274nm ~
199% I I 1 199% 246nm
G> ~ en ., c c 80% 8. 89% 8. en II) ..~ ..~ .. 3... s... !l !l ~ 69% ~ 60% "'0 "'0 ...~ ...Ill c c Ill Ill ... ~ II) ..~ ..~ 0 0 :s 40% :s 40% ii; ~
29% 20%
9%;----+----~--~--~----+----t----~--~~~----+ 0%;---~~--~----+----+----+----+----+---~ 200 2~9 390 3~9 499 4~9 n• 359 499 4~9 ~00 5~9 nrn
Wavelength Wavelength
FIG. 3.4 (a) The fluorescence excitation scan of zearalenone [3 J.Lg ml-i] in methanol:water (7:3) and (b) the fluorescence emission scan of the 0\ -..,J same solution at an excitation wavelength of 274 nm .68
·3.1.8 Nuclear magnetic resonance and mass spectrometry tech niques
The identity of two of the toxin standards (neosolaniol mono
acetate and fumonisin B1 ), which had previously been isolated within the Institute, were ·confirmed by MS and 1H and 13c N.M.R. spectroscopy, by Dr R Vleggaar, CSIR, Pretoria. The results for the toxins were compared with those previously published (Cole and Cox, 1981; Bezuidenhout et al., 1988).
3.1.9. Analytical standards - conclusion
Using the techniques described in ~ections 3.1.2 3.1.8, the identities of the analytical standards were confirmed. The absence of contaminants, as determined using a number of methods, in each of the standards analysed, appeared to indi cate a high degree of purity. Some of the physico-chemical data generated for the standards used in this study will be . incorporated into an ongoing IUPAC Food Commission project dealing with the provision of such data on mycotoxin stan- dards.
3 • 2 COMPARISON OF TYPE A TRICHOTHECENE METHODOLOGY.
3.2.1 Extraction procedures
The extraction procedures of the two methods evaluated, though similar, differed with regard to the primary purifica- 69
tion step. While the first method (Cohen and Lapointe, 1984)
used solvent partitioning of the toxin from an aqueous phase into ethyl acetate the latter (Sydenham and Thiel, 1987) made use of a hydrophilic (Kieselguhr) matrix. Use of the
Kieselguhr column resulted in.a partially purified eluate free from emulsions, a problem often encountered with solvent-solvent partitioning procedures. Both methods used a short silica gel column for the secondary purification step, however, in addition, the method of Cohen and Lapointe, (1984) also used a "cyano" column as a tertiary sample preparation step. Difficulty in obtaining the specified cyano column meant that a similar substitute column (cyano Bond Elut) had to be used. Both sample preparation methods were
found to be rapid and simpl~.
· 3.2.2 Chromatographic procedures
Greater sensitivity was achieved following the formation of the heptafluorobutyryl derivatives of the type A trichothe cenes than their corresponding trimethylsilyl derivatives. The method of Cohen and Lapointe, (1984) employed a splitless injection of a preformed purified sample derivative and se paration of the extract on a DB-5 capillary column. The re
tention times for diacetoxyscirpenol and T-2 toxin (using the conditions outlined by Cohen and Lapointe, 1984) were in the order of 41 and 50 minutes, respectively. The method of Sy denham and Thiel, (1987) used a split injection technique and separation of similarly prepared derivatives on a slightly 70
Q) 1/) c 0 a. 1/) Q) 0::
L...... 0 0 Q) b -Q) 0
Q) L...... :J a. a c 0 u c 0 ....L.. ~ f.s.d. 0 Q) w
0 5 10 15 TIME (minutes)
FIG. 3.5 Chromatogram of a spiked control maize sample ex- tract (prepared according to the method of Syden ham and Thiel, 1987) showing the chromatographic positions of (a) diacetoxyscirpenol, (b) neosola- niol monoacetate and (c) T-2 toxin, as their res- pective heptafluorobutyrate esters more polar phase column (Table 2.1), giving retention times of approximately 10 and 15 minutes for diacetoxyscirpenol and T-2 toxin, respectively. Although the introduction of sample extract via a splitter, onto a chromatographic column, can 71
result in reproducibility problems, the use of neosolaniol
monoacetate as an internal standard in the method (Sydenham and Thiel, 1987) reduced these errors to a minimum {Table 3.2). The suitability of neosolaniol monoacetate as an inter- nal standard could be attributed to the fact that it is also a type A trichothecene (Table 1.1), with a structure and chemical properties similar to those of diacetoxyscirpenol
and T-2 toxin. Using the prevailing chromatographic condi- tions, neosonaniol monoacetate eluted equidistant between
diacetoxyscirpenol and T-2 toxin (Figure 3.5).
TABLE 3.2 RECOVERIES OF DIACETOXYSCIRPENOL AND T-2 TOXIN (AND ESTIMATES OF ERROR1 ) USING THE PROCEDURE OF SYDENHAM AND THIEL, (1987)
Diacetoxyscirpenol T-2 toxin
Recovery (%) 88.2 88.0 Error (%) pooled within run 1.7 1.6 .cv (%) 1.9 1.8 Error (%) between run 4.7 4.1 cv (%) 5.3 4.7 Error (%) total 5.0 4.4
1 Estimates of error were calculated according to Krouwer and Rabinowitz, {1984). Data were compiled from 10 separate ex-
tractions with triplicate determinations per extraction, on -1 control maize spiked at a level of 1200 ng g diacetoxy- scirpenol and T-2 toxin 72
3.2.3 Toxin recoveries and method detection limit
Using the chromatographic conditions outlined in Table 2.1, the two methods were evaluated by analysing control maize to -1 . which 1200 ng g of each target toxin had been added. Tripli- cate recoveries using the method of Cohen and Lapointe, (1984) averaged 82.3 and 79.4% for diacetoxyscirpenol and T-2 toxin, respectively, and were within the range of recoveries repor- ted by the authors. The recoveries of diacetoxyscirpenol and T-2 toxin using the method of Sydenham and Thiel, (1987) were found to be slightly higher at 88% and 88.2%, respectively (Table 3.2). The total errors (Table 3.2) for the· latter method were extremely low, with the major source emanating from the "between-run" factors. The detection limit of each -1 method was approximately 100 ng g (parts per billion), as determined using spiked control maize. Each method was also assessed by analysing a fungal culture (F. armeniacum MRC -1 4481 - section 2.1.2) previously found to produce 440 ng g T-2 toxin. Figure 3.6.1 shows the chromatogram obtained from the fungal culture extract, following purification procedures according to Sydenham and Thiel, (1987). Diacetoxyscirpenol was not detected in the extract, however a well resolved peak corresponding to T-2 toxin, eluted after approximately 15 minutes. These observations were verified by spiking the sample with a similarly derivatized T-2 toxin standard. The chromatogram illustrates that T-2 toxin can be successfully extracted and identified from a fungal culture matrix, at re- latively low concentrations, without undue interference from co-eluting contaminants. Figure 3.6.2 shows the chromatogram 73
obtained from the same culture, following the procedure des-
eribed by Cohen and Lapointe, (1984). T-2 toxin was not re- covered, however, the use of non-specified materials during the sample purification stage of the method may have contri buted to this observation.
2
Q,) fn ~ 0 Q,. fn Q,) Soc Soc b ....0 CJ Q,) 1 ....Q,) "d Q,) b 3Q,. a c ~ CJ c e~ I ....CJ ~ ~ ' 0' 5 10 15 0 5 10 15
Time
FIG. 3.6 Chromatograms of the purified extracts of a maize
culture (MRC 4481) analysed according to the methods of (1) Sydenham and Thiel, 1987 and (2) Cohen and Lapointe, 1984. The alphabetical nota- tions a, b and c correspond to the chromatographic positions of diacetoxyscirpenol, neoso1aniol mono- acetate and T-2 toxin, respectively 74
3.2.4 Type A trichothecene methodology - summary
When analysing control maize, the methods compared favou
rably. However, the use of neosolaniol monoacetate as an in
ternal standard in the method of Sydenham and Thiel, (1987) contributed significantly to the quality of the recorded data
and resulted in the elution of the toxins within 15 minutes.
It should be noted that the method would also be suitable for
the extraction and chemical determination of neosolaniol mo
noacetate itself. This was demonstrated in a recent publica
tion (Sydenham et al., 1989c) on the production of neosola niol monoacetate by an undescribed Fusarium species where, with minor modifications, the same method was used.
Neosolaniol monoacetate has never been reported as a natural contaminant of grain samples and is itself a rare trichothe
cene, having been isolated only from three Fusarium species
and four Russian isolates belonging to the section Sporotri
chiella (Sydenham et al., 1989c).
3.3 COMPARISON OF TYPE B TRICHOTHECENE METHODOLOGY
3.3.1 Extraction procedures
The extraction and purification procedures of the two methods eyaluated, differed considerably. According to the publica tion of Tanaka and co-workers (Tanaka et al., 1985a), the use of acetonitrile/water (3:1) as an extraction solvent, resul- 75
ted in superior deoxynivalenol and nivalenol recoveries than other solvent blends exa~ined (including several methanol/ water blends). The authors also reported that the acetonitri le/water (3:1) solvent blend resulted in extracts containing fewer "interfering components" than the corresponding metha nol/water blends. The method subsequently involved the remo val of some interferences via· solvent partitioning with hexane followed by two separate chromatographic purification steps on Florisil and disposable silica gel columns, respec tively. Although this method was found to be time consuming (especially at the Florisil column stage), using relatively large volumes of organic solvents, it did result in a puri fied extract containing the two target toxins.
The second method evaluated (Scott et al., 1986), which is a modification of an earlier methbd (Scott et al., 1981), in volved extraction with methanol/water, precipitation with ammonium sulphate solution, followed by partitioning of the toxins on a hydrophylic matrix and final purification on a short silica gel chromatographic column. This method was found to be relatively rapid, using minimal volumes of sol vents and resulted in a visibly "less contaminated" extract than that resulting from the sample preparation procedure methoq of Tanaka et al. (1985a).
3.3.2 Chromatographic procedures
Although it is known that packed column GC inherently allows 76 for maximum column capacity (ie the highest sample equivalent weight that may be applied to an analytical GC column), the superior resolution attainable on capillary GC columns make their use highly desirable, especially when confronted with complex matrices. Therefore, extracts prepared by both analy tical procedures were analysed by capillary GC, even though one of the methods (Tanaka et al., 1985a) specified packed column GC as the chromatographic separation technique.
Both analytical procedures required the formation of trime thylsilyl derivatives of the purified sample extracts prior to chromatographic separation. Scott (Scott et al., 1986) suggested that the preparation and separation of HFBI deriva tives of samples (as used for the type A trichothecenes) could be used for confirmation purposes. Tri-Sil TBT was used as the silylation reagent (Scott et al., 1986) and it was found to form stable derivatives within a short reaction time (15 minutes), which exhibited excellent electron capturing characteristics and resulted in a far less complex capillary chromatogram than that obtained from a corresponding HFBI derivative. Scott et al. (1986) reported that the chromatographic separation of the nivalenol heptafluorobutyrate resulted in the appearance of two peaks on certain specified packed GC columns, as opposed to a single peak on capillary columns.
Figure 3.7.1 shows ·the chromatogram (corresponding to a 100
~g mass sample weight) obtained from a control maize sample spiked with 1000 ng g-1 deoxynivalenol and nivalenol (prior 77
to extraction), prepared by the method of Tanaka et al.
(1985a). Figure 3.7.2 shows a similar extract prepared accor ding to the method of Scott et al. (1986). In each case, fusarenon-X (a type B trichothecene similar in both chemical
2 1 •
1 • 3 •
2 2 • 1 • 3 •
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Time
FIG. 3.7 Chromatograms of control maize spiked with deoxyni valenol and nivalenol (at 1000 ng g-1 ) and extrac- ted according to the methods of (1) Tanaka et al. 1985a and (2) Scott et al. 1986. The numerical no- tations 1, 2 ·and 3 correspond to the chromatograp hic positions of deoxynivalenol, fusarenon-X (in ternal standard) and nivalenol, respectively 78
structure and characteristics to deoxynivalenol and nivale
nol - Table 1.2) as its similarly prepared trimethylsilyl de-
rivative (designated peak #2), was used as an internal ~tan- dard. Fusarenon-X was found to elute equidistant between deoxynivalenol and niva1enol (designated peaks #1 and #3, respectively). The major chromatographic difference between the two methods was the presence of a major component eluting at ± 17.5 minutes (Figure 3.7.1) in the extract prepared ac cording to the method of Tanaka et al. (1985a). However, the presence of the peak did not compromise the resolution of the deoxynivalenol, fusarenon-X and nivalenol peaks, which eluted at ± 22.6, 24.1 and 25.1 minutes, respectively.
3.3.3 Toxin recoveries, reproducibility and method detection limit
Using the chromatographic conditions outlined in Table 2.2, control maize extracts spiked with deoxynivalenol and nivale nol ·(at 1000 ng g-1 each toxin) were analysed in triplicate. Recoveries, using the method of Tanaka et al. (1985a) were found to average 89% (SD = 2.6) and 83% (SD = 2.5) for deoxy nivalenol and nivaleno1, respectively. Slightly higher ave rage recoveries of 92% (SD = 1.3) and 88% (1.6~) for deoxyni valenol and nivalenol, respectively, were obtained using the procedure of Scott et al., (1986).
The detection limit of each method was found to be approxi mately 40 ng g-1 for each toxin, using the prevailing chroma- 79
tographic conditions. This compared with the detection limits -1 -1 of 20 ng g and 2 ng g , reported by Scott et al. (1986)
and Tanaka et al. (1985a), respectively. Differences in the
chromatographic parameters used by the authors as opposed to
those used in this study may have contributed to these appa-
rent anomalies with regard to method sensitivity.
The use of fusarenon-X as an internal standard improved the
reproducibility as shown by the standard deviations calcula-
ted on 6 injections each of derivatized control maize extracts including and excluding fusarenon X (SD = 7.8 with- out internal standard reduced to 0.83 with internal stan- dard). The use of an internal standard was necessary since a manual splitless injection system was used. However, had an automatic injection system been available, the ,reproducibili- ty of injection would have been even better, and would no doubt have been reflected in lower standard deviation values than those reported above.
3.3.4 Type B trichothecenes - summary
The presence of deoxynivalenol as a major, contaminant of head blighted or "scabby" wheat has been reported in many coun- tries worldwide. Its detection in grains from Canada and the united states (Trenholm et al., 1981, 1983; Romer, 1983;
Scott, 1983; Scott et al., 1984; Eppley et al., 1984; Hagler et al., 1984; Seitz and Bechtel, 1985; Shotwell et al., 1985) gave rise to the rapid development of methods for its deter- 80 mination, predominantly by scientists in those countries.
Conversely, nivalenol (the toxicity of which is higher than that of deoxynivalenol) appears to be the major contaminant of Japanese grains, prompting extensive method developments for the co-determination of deoxynivalenol and nivalenol in grains, by Japanese scientists (Kamimura et al., 1978, 1981; Yoshizawa, 1984; Tanaka, 1985a). Poor recoveries of nivalenol inherent in the methods used by canadian scientists have sirice been rectified and Scott et al. (1986) and Lauren and
Greenhalgh,· (1987), ha~e reported excellent recoveries for nivalenol. The latter method (Lauren and Greenhalgh, 1987) is a HPLC procedure, however, since capillary GC was the preferred chromatographic technique (because of its superior resolution over HPLC), the method was not included for evaluation purposes.
A brief attempt to further reduce the sample preparation time of the method according to Scott et al. (1986), by replacing the short silica gel column with a similarly packed disposa ble cartridge, resulted in unacceptably lower recoveries of both target toxins (approximately 10% reduction) . In summary'· of the two analytical methods evaluated, the speed of sample preparation coupled with the excellent recoveries of both target toxins, made the method of scott et al. (1986) prefe rable. 81
3.4 COMPARISON OF ZEARALENONE METHODOLOGY
3.4.1 Extraction procedure
Subsamples of control maize were extracted according to the three procedures outlined in sections 2.5.1.1 - 2.5.1.3, which differed con~iderably with respect to their methodolo gies. The AOAC method (1984) involved extraction of the maize·with chloroform/water followed by chromatographic sepa ration from lipid-type material on a silica gel column, using large volumes of organic solvents. Final purification invol ved partitioning of the target toxin into acetonitrile. Al though the method specified TLC as the chromatographic sepa ratory technique, an aliquot of the purified extract was separated by HPLC and monitored by fluorescence detection (due to the increased resolution and sensitivity of the tech nique when compared to TLC). The use of a large chromatograp hic column (silica gel step), used in the AOAC procedure, resulted in this method being the most time consuming of the three·evaluated. The most rapid sample preparation method was found to be that of Chang and DeVries, (1984) which involved extraction with dichloromethanejwater, precipitation of pro teinaceous material with cupric carbonate and partitioning of the zearalenone into acetonitrile. The third method (Bagneris et al., 1986), which is a modification of that adopted by the
AOAC, following successful collaborative studie~ reported by
Bennett et al. (1985), involved extraction with chloroform/ water followed by primary partitioning into a base solution 82
and secondary partitioning into dichloromethane. The sample
extracts prepared by the latter method were visibly the least
complex, being colourless or slightly opaque, whilst those extracts prepared by the previous methods were yellow in colour.
3.4.2 Chromatographic procedures
Although the presence of z_earalenone in sample extracts may be monitored by HPLC using UV detection at several different
wavelengths (Figure 3.3), fluorescence detection affords greater sensitivity and selectivity for the determination of zearalenone. Sample extracts were, therefore, monitored by fluorescence detection, using the excitation and emission wavelengths as determined in section 3.1.7 (Figure 3.4). The literature concerning the choice of fluorescence detector
wavelengths for the determination of zearalenone is, however, ··somewhat confusing. Figure 3.8 (a-e) shows a series of chro matograms of a single sample extract screened under identical chromatographic conditions and instrument sensitivity set tings, but monitored at a range of wavelengths cited in the literature (Sydenham et al., 1988). With all the combina tions of excitation and emission wavelengths (Figure 3.8 a-e), both zearalenone and alternariol monomethyl ether {AME an Alternaria mycotoxin present in the samples screened, but which itself does not· form part of this study) could be de tected. The choice of excitation and emission wavelengths substantially influenced the sensitivity of detection. In 83
this study, the recorded fluorescence wavelength maxima (Figure 3.4) were similar to those previously published (Chang and DeVries, 1984).
b c d e •
•
• 0 6 12 0 6 12 0 6 12 0 6 12 0 6 12 TIME (minutes l FIG. 3.8 Chromatograms of a sample extract monitored at va- rious wavelengths (excitation wavelength (nm), emi~sion wavelength (nm), reference): (a) 236i 418, Trenholm et al., 1984; (b) 274, 418, Bennett et al., 1985, Bagneris et al., 1986; (c) 280, 465, chang and DeVries, 1985; (d) 320, 445, Sydenham et al., 1988; (e) 335, 404, Sydenham et al., 1988. Key: e , zearalenone; T alternariol monomethyl ether 84 Figure 3.9 shows the chromatogram of an unspiked control maize sample, prepared according to the AOAC method (1984), using chromatographic conditions similar to those cited in -1 Table 2.3 (except that a flow rate of 0.5 ml minute was .. used). The chromatogram res presents a 120 mg sample equiva- lent weight injected onto the chromatographic column. Whilst the chromatographic position of zearalenone (approximately 9 minutes - designated by an asterisk) was clearly contaminant - free, major component peaks were found to elute after 18 and 43 minutes. 0 10 20 30 40 50 Time (mins) FIG. 3.9 Chromatogram of an unspiked control maize sample extracted according to the method of the AOAC, 1984. The chromatographic position of zearalenone is designated by the asterisk 85 Figures 3.10.1 and 3.10.2, show the chromatograms obtained from similar extracts prepared according to the methods of Bagneris et al. (1986) and Chang and DeVries, (1984), respec- tively. Using these.methods, all components were found to elute from the analytical column within 12 minutes. No conta minant peaks were observed in the chromatographic position of zearalenone in the extract prepared by the method of Bagneris et al. (1986). However, minor contamination could be seen in the same chromatographic area in the latter extract (Chang and DeVries, 1984; Figure 3.10.2). 3.4.3 Toxin recoveries and method detection limit Subsamples of control maize spiked with authentic zearalenone . -1 · standard (at 150 ng g ) were prepared and used to determine toxin recoveries (in triplicate), for the three methods. Ave- rage recoveries and coefficients of variation (given in brae- kets) of 84% (± 8.7), 82% (± 7.2) and 78% (± 6.0) were re corded for the methods according to the AOAC (1984), Bagneris et al. (1986) and Chang and DeVries, (1984), respectively. Using control maize, and in the absence of interfering com- pounds, the detection limit of the methods was found to be in the order of 10-20· ng g-1 3.4.4 Zearalenone methodology - summary ...;1 . Although, at a spiking level of 150 ng g 1n control maize, the AOAC method (1984) gave the highest toxin recovery, a 86 1 2 0 1 0 4 8 12 0 4 8 12 Time (mins) FIG. 3.10 Chromatograms of unspiked control maize extracted according to the methods of (1) Bagneris et al., 1986 ~nd (2) Chang and DeVries, 1984. The chromato- graphic position of zearalenone is designated by the asterisk number of inherent method characteristics, including sample preparation time, the use of large volumes of organic sol- vents (including the carcinogen benzene) and the necessity for prolonged analysis times (to elute all components), com bined to make the method the least desirable. The method of 87 Bagneris et al. (1986) gave the second highest.toxin recovery, whilst it also gave both visibly and chromatographically, the ' least complex sample extract of the three methods. Sydenham and co-workers (Sydenham et al., 1988) also reported that the three methods evaluated in this study also extracted a totally unrelated toxin (ie AME), from a series of mixed feed samples. The limitations of TLC as a separatory technique was also demonstrated (Sydenham · et al., 1988), since the zearalenone and AME co-contaminants of the feed samples could not be fully resolved by TLC, but were so by HPLC. The impor tance of the choice of excitation and emission wavelengths in fluorescence detection were demonstrated in the same publica- tion. 3.5 COMPARISON OF MONILIFORMIN METHODOLOGY ' 3.5.1 Extraction procedures The method of Thiel et al. (1982b) was found to be an extre mely simple three step procedure, consisting of (a) extrac tion of the toxin fro~ the matrix with distilled water fol lowed by (b) centrifugation and (c) supernatant filtration prior to HPLC separation. Although the method required minimal sample handling, the· major time consuming step was the one hour required for toxin extraction. The method of Scott and Lawrence, (1987) appeared to be far more involved, consisting of extraction with acetonitrile/water, a defatting process · via solvent partitioning with hexane and two succes- 88 sive cleanup stages on small c18 and neutral alumina columns, respectively. However,·not only was the extraction procedure found to be more rapid than that of Thiel et al. (1982b), it also resulted in an extract that was visibly less contamina ted than the opaque extract resulting from the method of Thiel et al. (1982b). 3.5.2 Chromatographic procedures The UV spectrum of moniliformin (Figure 3.2), clearly identi fies 'two absorption maxima that might be used for monitoring the presence of moniliformin (229 nm and 254 nm). When moni toring moniliformin, the 229 nm wavelength was found to give . a threefold increase in sensitivity over the 254 nm wave length. However, the two wavelengths could be used in con junction with each other for confirmation purposes, where the maximum sensitivity was not a prerequisite (Figure 4.6). Paired ion chromatography was used by both methods for the separation of the purified extracts, however Thiel et al. (1982b) also used ion exchange chromatography for confirmato ry purposes. Using the specified HPLC conditions the monili formin peak eluted at a retention time of approximately 8 minutes. Under the specified analytical conditions, the greatest dif ference between the methods was the sample equivalent weight applied to the analytical column (ie 1.5 mg for the method of Thiel et al. 1982b, as opposed to 25 mg for the method of 89 Scott and Lawrence, 1987). To a large extent the attainable sensitivity of any analytical method is dependant ·upon the efficiency of the sample cleanup, since this is the limiting factor with respect to the mass equivalent of matrix that can 1 2 0 4 8 12 16 0 4 8 12 16 Time (minutes) • FIG. 3.11 Chromatograms representing 17 mg mass sample equi valents injected onto the analytical columti, of unspiked control maize extracted according to the methods of (1) Thiel et al., 1982 and (2) Scott and Lawrence, 1987. The chromatographic position of moniliformin, in both chromatograms, is desig- nated by the asterisk 90 be injected onto any chromatographic column. The methods of Thiel et al. (1982b) and Scott and Lawrence (1987), were therefore altered slightly so as to compare the chromatograms obtained from each method, following the injection of the same sample mass fraction onto the HPLC column. Figures 3.11.1 and 3.11.2 illustrate a 17 mg sample equivalent weight extracted from a control maize sample, according to the methods of Thiel et al. (1982b) and Scott and Lawrence, (1987), respectively. The figures clearly demonstrate the degree of contamination present in the extract prepared according to the me~hod of Thiel et ~1. (1982b Figure 3.11.1), whilst the absence of this contamination in the extract prepared according to the method of Scott and Lawrence, (1987), is similarly illustrated. In both cases, the chromatographic position of moniliformin is indicated by an asterisk. 3.5.3 Toxin recoveries and method detection limit Taking into account the fact .that different sample equivalent weights are injected onto the column, different spiking le~ vels were necessary for the two methods. Two subsamples of control ·maize, spiked with 50 ~nd 3 IJ.g g-1 moniliformin, res pectively, were prepared, extracted and analysed in tripli- cate according to the methods of Thiel et al. (1982b) and I Scott and Lawrence, (1987) - average recoveries of 80% and 96% ~ere obtained, respectively. The detection limit (using. control maize and the chromatographic conditions outlined in Table 2.4) for the method of ·Thiel et al. (l982b) was found 91 -1 to be in the order ,of 10 JJ.g g , whereas the detection limit for the method of Scott and Lawrence, (1987) was found to be -1 200 ng g By altering the detector sensitivity, the detection limits (using control maize) could be reduced by' factors of approximately 2.5 and 10, respectively. This dif ference was due primarily to excessive matrix interferences encountered in the method according to Thiel et al. (1982b), which were substantially reduced in the method of Scott and Lawrence, ( 1987). 3.5.4 Moniliformin methodology - summary The overwhelming supe~iority of the method according to Scott and Lawrence, (1987) over that according to Thiel et al. (1982b), was demonstrated by a number of observations. The former method not only gave a less complex extract, it also exhibited a higher target toxin recovery and greater sensiti vity, than the latter method. A third method for the determi nation of moniliformin in maize was initially considered for evaluation purposes (Shepherd and Gilbert, 1986). Problems were encountered with regard to toxin recovery at the secon dary cleanup stage of the method. Due to the non-availability of some of the specified materials, alternatives were used. These alterations to the method may well have been the cause of the low recoveries encountered. Due to these problems, it was not possible to fully evaluate the method, and therefore, for the purposes of this study it was excluded from the eva luation procedure. 92 8: INNOVATIVE METHODOLOGY 3.6 TLC DETERMINATION OF_THE.FUMONISINS TLC has been used as a monitoring system for the isolation of fumonisins B1 and B2 from culture material of F. moniliforme MRC 826 (Gelderblom et al., 1988). The fumonisins could be detected using silica gel 60 TLC plates developed in chloro form :methanol : acetic acid (6:3:1), followed by spraying with p-anisaldehyde solution and visual inspection (as des cribed in section 3.1.2). However, as the limit of detection (for the TLC method) was found to be in the order of 500 ~g 1 g- for each fumonisin toxin, it was considered unsuitable as a technique for the determination of the fumonisins (FB 1 -' or FB2 ) in naturally contaminateq samples. 3.7 HPLC DETERMINATION OF FUMONISIR 81 (AS ITS MALEYL DERI VATIVE) Siler and Gilchrist, (1982) used a derivatization procedure for the determination of a host-selective phytotoxin (TA toxin), produced by Alternaria alternata f. sp. lycopersi ci, by the introduction of the UV absorbing maleyl group onto the primary amine present in the structure of the toxin. Figures 3.12.1 and 3.12.2 illustrate the structures of TA toxin and fumonisin B1 , · respectively, and clearly show the structural similarity.of the two compounds which are produced by separate, totally unrelated genera of fungi. 93 1 . CH3 CH I I 3 CHa-CHa-CH-c;:H-CH-CH2-CH-CH2-{CH2h-CH-CH-CH2-CH-CHa-NH1 I I · · 4 I I I OH 0 ' OH OH OH I C=O I ~H 2 ~ CH-C-OH I CH -C-OH 2 n 0 2 c~ c~ · ~~I I. . CH 1-.CHa13 CH-JH-I(H-CH 2-CH-CH 2-lH-{CH 2};-lH-CH 2-1H-~H-CH 1 y ? OH OH OH NH 2 . "/ O=C C=O I I W CfHa ~Ha ~ HO-C-CH CH-C-OH I I HO-C-CH2 CHz-C-OH ~ & ·FIG. 3.12 Chemical structures of· (1) the Alternaria phyto..:. toxin TA toxin and (2) the Fusarium mycotoxin fumonisin B1 The difference between the two toxins is the absence of one TCA group and a shorter aliphatic aminopentol moiety in c3 the structure of the Al toxin. Due to this structural simila- rity, the derivatization procedure (Siler and Gilchrist, 1982) was, with minor modifications, applied to a fumonisin B1 standard. Figures 3.13.1 ~nd 3.13.2 show the chromatograms obtained from a derivative blank and a fumonisin B1 standard, respectively. The derivatized fumonisin B eluted at a reten- 1 94 tion time of± 6.0 minutes (Figure 3.12.2). Using control 1 maize spiked with fumonisin B1 (at 100 ~g g- ), triplicate recoveries were performed. The average cor~ected recovery was found to be 66.3% (SO = 1.48%) and the limit of detection was -1 in the order of 10 ~g g 1. 2 .015 i Q M= .010 t! Q) Ul 0 =g.. fll Q) loc loc FB1 0 ...u ~ .005 + .,Q,) ~ 0 0 2 4 6 8 10 0 2 4 6 8 10 Time FIG. 3.13 HPLC chromatograms of (1) a maleyl derivative blank and (2) a similarly derivatized fumonisin B1 standard 95 The extraction, derivatization and chromatographic procedures for the determination of furnonisin B1 (as its rnaleyl deriva tive) in maize (sections 2.7.1, 2.7.2 and 2.7.6), were applied to two samples of control maize, three subsarnples of horne grown Transkeian maize (detailed in section 2.1.3).and a sam- ple of culture material (F. moniliforme MRC 826 - section 2.1.2). The results are given in Table 3.3.; 1 2 0.009 - 0.000 0 2 4 6 8 10 0 2 4 6 8 10 Time FIG. 3.14 HPLC chrdrnatograrn of (1) the rnaleyl derivative of sample M-84 and (2) the chromatogram of a similar- ly prepared extract of hand-selected maize kernels, showing the presence of a contaminant peak ( * ), which elutes in the chromatographic position of 96 Figure 3.14.1. shows the chromatogram obtained from a 5 ~1 aliquot of the maleyl derivative of an extract of the Trans- keian maize sample M-84. Baseline separation of the fumonisin B1 peak was not achieved, due to the presence of matrix in terferences from the maize substrate. Resolvation problems were encountered (especially when analysing naturally conta- minated maiz.e samples), following the chloroform partitioning step during the primary clean-up stage of the extraction pro cedure. The residue could not be fully dissolved in methanol: water (1:3). A gelatinous methanol soluble precipitate was normally encountered. Figure 3.14.2 details the chromatogram obtained from a simi- larly derivatized extract of a control maize sample. Whilst the degree of matrix interference is much lower than that ob- served in Figure 3.14.1, a small peak could be seen eluting at the chromatographic position of fumonisin B1 (correspon 1 ding to <10 ~g g- fumonisin B1 - Table 3.3). A similar peak was also obsel;'ved in a sample of commercially available maize meal and subsequently in 3 other maize samples. The presence of this peak in several maize samples indicated the possibi lity that it was an interfering compound found intrinsically in maize. Figure 3.15 illustrates the chromatogram obtained from an ex- tract of a fungalculture (F. moniliforme MRC 826) and clearly shows the reduced matrix interferences encountered chromatographically (when compared to Figure 3.14.1). 97 TABLE 3.3 LEVELS OF FUMONISIN B (FB ) AND TRICARBALLYLIC ACID (TCA) ASSAYED 1 1 AS THEIR MALEYL AND BUTYL ESTER DERIVATIVES, RESPECTIVELY! 2 Sample FB measured TCA calculated as TCA measured as the 1 present due to FB butyl ester deriva-, as the maleyl 1 . . 3 . 4 d er~vat~ve (maleyl derivative t~ve results) -1 -1 -1 (JLg g ) (JLg g ) (JLg g ) Hand selected 5 control corn <10 <4.9 <0.5 Commercial corn meal <10 <4.9 13 M-84/C <10 <4.9 21 M-84 44 22 101 M-84/F 83 41 164 MRC 826 9 280 4 530 6 400 1 Results currently in press (Sydenham et al., 1989a) 2 M-84/C Healthy maize kernels M-84 Mouldy maize kernels M-84/F = Fusarium-infected maize kernels MRC 826 = Culture mat.erial of Fusarium moniliforme MRC 826 3 -1 Determined by HPLCjUV; detection limit approximately lOJLg g 4 -1 Determined by GC/FID; detection limit = 0.5 JLg g 5 Subsequent analysis indicated that this was a contaminant peak eluting in the position of FB 1 98 Figure 3.15 also shows that fumonisin B can be monitored by 2 the same derivatization technique . .015 ..FB 1 -9 Q e(") .010 ~ Cl) =a..0 Cl) ...f 0 ....u ~ ~ .005 "tS- ~ ..FB 2 0 0 4 8 12 16 20 24 28 Time FIG. 3.15 Chromatogram of the maleyl derivative of an ex- tract of F. moniliforme (MRC 826), showing the presence of fumonisin B1 (rt = 6.0 minutes) and fumonisin B2 (rt = 24.0 minutes) Therefore, a number of matrix related factors demonstrated the limitations of the maleyl derivative procedure for the 99 .determination of fumonisin B1 in naturally contaminated maize samples. However, when analysing culture samples, matrix interferences, at both the extraction and chromatographic stages of the method, were substantially reduced. This may be explained by the fact that due to the far higher concentra tions of the fumonisins present in fungal culture samples, a far lower sample equivalent weight is injected onto the HPLC column (up to 50 times less)., significantly reducing matrix effects. 3.8 HPLC DETERMINATION OF FUMONISIN B1 (AS ITS FLUORESCAMINE DERIVATIVE) It was necessary to consider the use of detection techniques other than uv, for the determination of fumonisin B1 in natu rally contaminated maize, due to the relatively poor detec tion limit (10 ~g g-1 ) obtained by the maleyl derivatizationj UV procedure. Fluorescence detection tends to offer superior sensitivity and selectivity than uv for numerous compounds. In the case of fumonisin B1 it was necessary to introduce a fluorescent species (or label), due to the molecules' lack of intrinsic fluorescence characteristics. A number of fluores- cent labels are commercially available and several have been used extensively for the derivatization of primary amines, especially in the area of amino acid analyses (Perrett, 1985; Rosenthal, 1985). Dansyl chloride (Dns-C1, 1-dimethylaminonaphthalene-5-sulpho- 100 nyl chloride), ortho-phthalaildehyde {OPA) and 7-fluoro-4- nitrobenzo-2-oxa-1,3-diazole (NBD-F) were three fluorescent labels which were initially evaluated. Derivatization of fumonisin a1 (with these reagents) required reactions to be carried out at various pH, reaction time and temperature settings. Initial HPLC separations of the preformed deriva- tives (using a Waters 4-j.l.m c18 column - as used for both zearalenone and moniliformin analyses) resulted in pronounced r chromatographic tailing effects. Fluorescamine (4-phenylspiro [furan-2(3H)yl-1-phthalan]-3,3' dione), another fluorescent label that has been used for the analyses of amino acids, was also evaluated. It is a compound which lacks intrinsic fluorescence (unlike other labels - ie NBD-F), but it reacts rapidly at room temperature with prima ry amines (t~ = 100-500 milliseconds- Perrett, 1985), to form stable fluorophors (Dns-·C1 and OPA derivatives tend to be unstable). A fluorescamine- fumonisin B1 derivative com plex was also found to exhibit chromatographic tailing. This effect was reduced, to a large exterit, by using a HPLC column having high carbon loading and extensive end-capping features. Control maize and the Transkeian maize subsamples were analy- sed according to the method outlined in section 2.7.6. Figure 3.16.1 shows the chromatogram of a fluorescamine derivatized fumonisin B standard (65 ng), where two well resolved peaks 1 may be seen at the retention times of 14.3 and 16.5 minutes, \ 101 respectively. Figure 3.16.2 illustrates the chromatogram obtained from a similarly derivatized extract of M-84/F and Figure 3.16.3, the same extract spiked with derivatized fumonisin a1 • 1 2 3 0 6 18 0 6 18 0 6 18 T~e FIG. 3.16 HPLC chromatograms of the fluorescamine deriva- tives of (1) 65 ng fumonisin a1 , (2) sample M-84/F and (3) the chromatogram of the same extract spiked with derivatized fumonisin a1 When used as part of a post column detection system, a single fluorophore is chromatographed. However, HPLC separation of a preformed fluorescamine - primary amine complex often results in a dual peak for primary amines, as was the case with fumo- 102 nisin B (Figure 3.16.1). The two peaks result from the acid 1 alcohol and lactone forms of the fluorescent complex which exhibit identical fluorescent characteristics (Rosenthal, 1985), as shown in Figure 3.17. 0 Chemical structure offluorescamine HPLC separation of preformed Ouorescamine-primary amine complex results in dual peaks for many compounds Acid alcohol Lactone o0~ c-c;:-R II I I HO(Jxy--o..,.co N - 0 These two complexes exhibit identical fluoreacence characteristics FIG. 3.17 Flow diagram of a preformed fluorescamine-primary amine complex which results 1n· the formation of 2 derivatives The clean-up and derivatization procedure gave excellent re- 103 coveries (in excess of 90%) for the determination of pure fumonisin B1 • However, when extracts of control maize spiked with fumonisin B1 were analysed, the observed recoveries were far lower(< 50%). The cause for the anomaly appeared to be the presence of other primary amine containing compounds in the control maize sample, which were also derivatized by the available fluorescamine. Methods to remove these interferen- ces andjor the utilization. of a post column fluorescamine derivatization procedure were investigated, but still require further optimization. Therefore, due to inherent matrix effects, the procedure could only provide qualitative data pertaining to the presence of fumonisin a1 in maize. 3.9 CAPILLARY GC OF FUMONISIN.B1 Fumonisin B1 is a large molecule, having a molecular weight of 721. Attempts were made to chromatograph the complete mo- lecule intact, which required derivatization of hydroxyl, carboxyl and primary amine groups. Several derivatization procedures (Blau and.King, 1977) were evaluated. These proce- dures including the use of diazomethane, HFBI, Tri Sil~TBT, N,O-bistrimethylsilylacetamide (BSA) and N-methyl-N-(tert- butyl-methylsilyl)-trifluoroacetamide (MTBSTFA). In each case, a fumonisin B1 and a derivative blank were prepared for comparative chromatographic analysis. Using these derivatiza~ tion procedures (or the prevailing chromatographic conditions) no chromatographic peaks could be detected. 104 Fumonisin B1 was subsequently subjected to a hydrolysis, es terification and acylation procedure used for the capillary GC separation of amino acids (sections 2.7.3 and 2.7.8, Laba- darios et al., 1984). Chromatographic separation of the deri vative (using the conditions outlined in Table 2~7) resulted in the elution of a single peak. This peak could not be de tected when the acylation step alone was performed on an ali quot of the fumonisin B1 hydrolysate. However, esterification of an aliquot of the same hydrolysate (without the subsequent acylation step), resulted in the detection of the same single chromatographic peak. These observations suggested that the hydrolysis had successfully hydrolysed the fumonisin B1 , but that in the process (or using either the derivatization or chromatographic parameters outlined) only the esterified tri carballylic acid (TCA) moiety of the fumonisin B could be 1 detected, and not the anticipated aminopentol moiety. This supposition was confirmed upon esterification and GC analysi~ of an authentic TCA standard, which resulted in a single peak being detected at a retention time identical to that obser ved for the esterified extract of the fumonisin B hydroly 1 sate. Using the prevailing chromatographic conditions (Table 2.7), the detection limit for TCA was found to be in the order of 500 ng g -1 . Using the extraction/hydrolysis and esterification procedure outlined in section 2.7.1, 2.7.3 and 2.7.8, the levels of the TCA moiety present in the two control maize, three Transkeian maize and fungal culture samples (used for the evaluation of 105 the HPLC-maleyl derivative procedure) were determined by ca- pillary GC (as its iso-butyl ester), and the results obtained are given in Table 3.3 1 TCA.. 2 TCA.. 0 2 4 6 8 10 12 14 16 18 20 22 Time FIG. 3.18 Capillary GC/FID chromatograms of (1) a hydroly sed esterified.extract of M-84/C showing a major peak eluting at 18.3 minutes (Tricarballylic acid TCA), and (2) the same sample extract spiked with esterified TCA 106 Figure 3.18.1 shows the capillary GC-FID chromatogram obtai ned from a hydrolysed-esterified extract of M-84/C, whilst Figure 3.18.2 shows the same extract spiked with a similarly esterified fumonisin B1 standard. A well resolved peak cor responding to the isobutyl ester of TCA eluted at a retention time of 18.3 minutes (Figures 3.18.1 and 3.18.2). The TCA contribution due to the fumonisin B1 (determined as its maleyl derivative), present in each sample, was calcula ted and the results are given in Table 3.3. The levels of TCA determined experimentally (by the GC-FID procedure) were in variably higher in each sample than the corresponding TCA content contributed by the fumonisin·B1 . In the case of the hand-selected control maize sample, the peak observed in the chromatographic position of fumonisin B , as its maleyl deri 1 vative (Figure 3.14.2) was undoubtedly a contaminant peak since no corresp6nding TCA was detected in the sample. Since a similar sized peak was also observed in the commercial maize meal and M-84/C s~mple extracts, the levels of TCA de termined experimentally in these samples could not be direct ly attributed to the presence of fumonisin B1 . The fact that the TCA levels in 5/6 samples were higher than could be explained by the presence of fumonisin B1 , was not unexpected, as it has already been shown that TCA containing compounds other than fumonisin B1 (such as fumonisin B2 ) are produced by F. moniliforme (Gelderblom et al., 1988). Ano ther TCA containing fumonisin (fumonisin B3 ) has also recent- 107 ly been isolated from F. moniliforme (Gelderblom - personal communication). In the case of the MRC 826 culture material, the fumonisin B1 accounted for >70% of the total TCA detected experimentally in the hydrolysed-esterified extract, which agrees with the observation that fumonisin B is the major 1 fumonisin produced by F. moniliforme in culture (Gelderblom et al., 1988). In samples M-84 and M-84/F, fumonisin B con 1 tributed less to the total TCA than in the culture material, implying that more of the other fumonisins were present in these samples. A paper detailing the natural occurrence of fumonisin B in 1 the Transkeian maize subsamples, based on the chromatographic evidence presented in sections 3.7 - 3.9 was recently accep ted f6r publication (Sydenham et al., 1989a) 3.10 SUMMARY OF FUMONSIN B1 METHODOLOGY The chromatographic investigations covered in sections 3.6 - 3.9 resulted in a number of observations listed below: 1. That TLC was an inappropriate analytical method other than as a screening procedure for the presence of the fumoni sins in culture material. 2. That the maleyl derivative method was useful as a quan titative procedure for the analysis of the fumonisins B1 and B2 in culture material, but that the detection limit (10 108 1 ~g g - ) was insufficient for the detection of low concentra tions that would be present in naturaily contaminated samples. 3. That the fluorescamine derivative-fluorescence detection procedure, whilst potentially allowing superior sensitivity, still required optimization with respect to potential sample matrix effects. 4. That at present, using the hydrolysis and subsequent de~ rivatization procedures outlined, only the TCA moiety pre•ent in the structures of the known fumonisins, can be detected. These points therefore suggested that until methods are avai lable for the determination of the separate fumonisins, the hydrolysis-esterification procedure for the determination of the TCA moiety, might be used as a chromatographic indicator of the potential extent of the total fumonisin contamination of naturally contaminated maize samples. 109 CHAPTER 4 APPLICATION OF FUSARIUM MYCOTOXIN METHODOLOGY TO A SERIES OF HOME-GROWN TRANSKEIAN MAIZE SAMPLES ASSOCIATED WITH HUMAN OESOPHAGEAL CANCER RISK 4.1 OESOPHAGEAL CANCER IN THE TRANSKEI It has for many years been suggested that an association exists in Africa, between the occurrence of human oesophageal cancer and maize cultivation (Cook, 1971). In Africa, oeso phageal cancer rate is highest in the south-western districts of the Transkei ( ie. Kentani - Figure 4 .1), whilst the rate in the north-eastern region (ie Bizana, Figure 4.1) is relative ly low (Rose, 1973, 1982; Rose and McGlashan, 1975; Jaskie wicz et al~, 1987). Maize is the staple diet in bdth regions. An integral part of an ongoing mycotoxicological investiga tion, concerning maize gathered from the two regions, has .been the implementation of comparative studies of the domi nant mycoflora of the maize. The most consistant difference in the mycoflora in the home-grown maize during each of 3 years, has been the significantly higher incidence of Fusa rium moniliforme Sheldon in maize produced in the high rate area (Marasas et al., 1981). This fungus has also been asso ciated with foodstuffs in the Lin Xian district in the Henan province of China, which is one of the highest oesophageal cancer risk areas in the world (Li et al., 1979, 1980; Yang, 1980). 110 Transkei Indian ocean FIG. 4.1 Map detailing the position of Transkei within southern Africa, illustrating the locations of the low (Bizana) and high (Kentani) oesophageal cancer risk regions Previous reports concerning the correlation of the presence of F. moniliforme and oesophageal cancer risk rates in Transkei, were based on the random sampling of households in the various geographical areas (Marasas et al., 1981), but no information was available concerning the incidence of oeso- phageal cancer.in the occupants of the households. In a subsequent study (Marasas et al., 1988a) the prevelance of human oesophageal cytological abnormalities were determi- 111 ned by means of brush biopsy capsules in adult occupants in each of 12 households in a low, an intermediate and a high oesophageal cancer rate area, in Transkei. Mild cellular chan ges (folic acid deficiency, atypia and mild dysplasia) and advanced changes (dysplasia and cancer), occurred more frequently in the occupants of households in high, than in intermediate or low oesophageal cancer rate areas. A study of the dominant fungi present in home-grown maize samples collected from each of the households included in the cytological screening survey, showed F. monili:forme was significantly higher in maize from cytologically "affected" households in the high oesophageal cancer rate area than from "unaffected" households in the low oesophageal cancer rate area (Marasas et al., 1988a). Therefore, whereas the pre viously established correlation was between F. monili:forme and oesophageal cancer rate, the results provided evidence for an association between the fungus and oesophageal cytological abnormalities in living individuals. 4.2 FUSARIUM MYCOTOXIN ANALYSES The last part of this investigation is therefore, the appli cation of the most suitable methodology for the determination of Fusarium mycotoxins (discussed in Chapter 3), to the home-grown maize samples which were collected from the house holds of individuals involved in the cytological survey de- . tailed in section 4.1 (Marasas et al., 1988a). 112 TABLE 4.1 METHODS USED FOR THE DETERMINATION OF FUSARIUM MYCOTOXINS IN HOME-GROWN TRANSKEIAN MAIZE SAMPLES Fusarium Separation and Extraction and purification mycotoxin Detection technique (References) Diacetoxyscirpenol Capillary GC-ECD Sydenham and Thiel, 1987 and T- 2 toxin Nivalenol and Capillary GC-ECD Scott et al., 1986 Deoxynivalenol Zearalenone HPLC-Fluorescence Bagneris et al., 1986 Moniliformin HPLC-UV Scott and Lawrence, 1987 Fumonisins Capillary GC-FID Sydenham et al., 1989a 113 TABLE 4.2 TRANSKEIAN HOME-GROWN MAIZE SAMPLES (1985), MYCOLOGICAL AND 1 CHEMICAL RESULTS 5 6 . 7 8 District .E. _E. .E. Mon Zea N~v Don 2 3 4 -1 -1 -1 -1 -1 rnon. sub. ~· p,g g p,g g p,g g p,g g jlg g BIZANA 1 67 5 12 0.64 0.79 1. 02 0.09 8.0 2 17 1 61 0.38 1. 78 4.09 0.40 5.0 3 22 0 82 0.89 1. 26 2.42 4.39 13.1 4 34 6 47 0.94 0.54 1. 87 7.73 6.3 5 43 1 61 1. 21 2.94 15.20 0.55 10.2 6 50 10 29 7.51 1. 74 SL10 SL 20.8 11 7 26 27 16 8.55 ND 0.97 0.05 19.2 8 35 9 0 1. 69 ND SL SL 16.9 9 20 15 25 2.51 0.31 3.82 12.10 14.3 10 54 26 14 11.57 1. 36 SL SL SL 11 32 8 9 4.97 ND 1. 53 0.32 16.5 12 14 13 63 1. 51 3.28 10.64 0.56 6.5 KENT ANI 1 51 6 1 1.80 ND 0.08 0.09 16.7 2 94 1 5 0.52 ND SL SL 46.2 3 66 4 33 0.65 1.43 3.39 0. 72 20.1 4 89 2 3 0.35 ND SL SL 57.4 5 51 11 0 0.48 ND 0.90 0.90 32.4 6 34 0 0 1.12 ND SL SL 31~6 7 72 8 1 0.56 ND 0.50 0.90 36.2 8 61 2 3 0.95 0.12 0.92 9.16 SL 9 76 3 12 1.04 0.15 0.75 0.14 24.5 10 38 8 36 1.17 2.39 8.73 1.00 25.8 11 87 3 1 0.48 ND 0.14 0.11 518.0 12 93 8 1 0.91 0.23 1. 59 0.25 46.1 114 KEY TO TABLE 4. 2 1 No diacetoxyscirpeno1 or T-2 toxin were detected in any of the maize -1 samples (detection limit = 100 ng g ) 2 E. mon. -Kernels infected with Fusarium moniliforme (%) (Marasas et al., 1988a) 3 E. sub. -Kernels infected with Fusarium subglutinans (%) (Marasas et al., 1988a) 4 E. gram. - Kernels infected with Fusarium graminearum (%) (Marasas et al., 1988a) 5 Mon - Moniliformin (detection limit = 200 ng g-l) 6 -1 Zea - Zearalenone (detection limit = 20 ng g ) 7 -1 Niv - Nivalenol (detection limit = 40 ng g ) 8 1 Don - Deoxyniva1enol (detection limit = 40 ng g- ) 9 -1 TCA - Tricarba11ylic acid (detection limit = 500 ng g ) lO SL - Samp 1 e 1 ost 11 · ND - Not detected (see relevant toxin detection limits above) 115 Maize samples (12 each) from the high (Kentani) and low (Bizana) oesophageal cancer rate areas were analysed for the presence of the Fusarium mycotoxins using the methods de- .tailed in Table 4.1. The previously reported mycological re- sults (Marasas et al., 1988a) and the generated toxin results for each of the 24 samples, are given in Table 4.2. 4.3 CHROMATOGRAPHIC ANALYSES AND CONFIRMATION 4.3.1 Type A trichothecene analyses The presence of diacetoxyscirpenol and T-2 toxin was not de- tected in any of the maize sample extracts (at a detection -1 limit of 100 ng g ). Figure 4.2.1 illustrates the capillary GC chromatogram obtained from a maize sample extract as its heptafluorobutyryl ester, containing neosolaniol monoacetate (as the internal standard). Figure 4.2.2 shows the chromato gram of the same extract spiked with similarly derivatized 'diacetoxyscirpenol and T-2 toxin standards. 4.3.2 Type B trichothecene analyses Nivalenol and deoxynivalenol were detected in each sample analysed, using a capillary GC-ECD procedure. Initial verifi- cation of the presence of deoxynivalenol and rtivalenol in the sample extracts, was by the spiking (and subsequent chromato- graphic separation) of the extracts with similarly derivati- zed standards. 116 2 2 2 \ \ \ t 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 Time FIG. 4.2 Capillary GC chromatograms of (1) an extract of a Transkeian maize sample as its heptafluorobutyryl ester and (2) the same extract spiked with the target type A trichothecenes. The numerical nota- tions 1, 2 and 3 correspond to the chromatographic positions of diacetoxyscirpenol, neosolaniol mono- acetate (internal standard) and T~2 toxin, respec- tively Due to problems with the cold storage facilities in which the maize samples were maintained, several of the· samples were severely damaged and subsequently excluded from the analyses. In order to verify the identity of nivalenol and deoxynivale- nol in a number of maize samples, extracts were analysed and monitored by capillary GC~Ms using the conditions outlined in section 2.8.2. 117 Figure 4.3.1 shows the capillary GC-MS chromatogram of a maize sample extract, as its TMS derivative (Bizana sample -1 number 9), previously shown to contain 12~1 ~g g deoxyniva- lenol (~able 4.2). The chromatographic position of deoxyniva lenol (DON) was found to be approximately 24.8 minutes. The mass spectrum (100-400 mjz) of this peak is shown in Figure 4.3.2. The excellent agreement between the retention times and the mass spectra of the DON peak in the maize sample and that of a similarly prepared authentic DON standard (as its corresponding TMS derivative- Figure 4.3.3), provided une quivicol evidence for the presence of deoxynivalenol in the maize extract. Similarly, Figure 4.4.1 shows the capillary GC-MS chromato- gram of Bizana sample number 5, which had been found to con tain 15.2 ~g g-1 nivalenol (NIV). The presence of nivalenol, in the sample, was confirmed via the agreement of the reten- tion times and mass spectra of the NIV peak in the sample and that obtained for authentic nivalenol (Figures 4.4.2 and 4.4.3, respectively). Several other sample extracts contaminated with various con- centrations of both Fusarium toxins were also analysed by capillary GC-MS, and in each case the retention times and mass spectra compared favourably. 118 ~DON 21. ,.. 411. -:!Ioiii -zso• -Boll KMTllli ••• t I .I ••• 11'11 2 •••• :IS .I •••• 1 •• ••• "'11•••• 3 :r.s .1 '·' - FIG. 4. 3 (1) Capillary GC-MS chromatogram of a sample ex tract (Bizana #9) as its trimethylsilyl (TMS) deri- vati ve, ( 2) mass spectrum of the deoxynivalenol (DON) peak (rt = 24.8 minutes) and (3) the mass spectrum of an authentic DON-TMS derivative 119 II, ~ NIV 11!11 -•••• llo• 1••• I ,I "·' II .I :H .I IC .1 1111 I 2 II. I '·' 1111 1•• 1 I .I II .I "·' 3 11.1 Ll FIG o 4 o 4 (1) capillary GC-MS chromatogram of a sample extract (Bizana #5) as its trimethylsilyl(TMS) derivative, (2) the mass spectrum of the nivalenol (NIV) peak (rt = 27o7 minutes) and (3) the mass spectrum of an authentic NIV-TMS derivative 120 4.3.3 Moniliformin analyses Moniliformin was detected in each sample analysed at levels -1 . of between 0.35 - 11.57 ~g g . F1gure 4.5.1 shows the HPLC chromatogram of Bizana sample number 11 which was found to cont~in 4.97 ~g g-1 moniliformin. Figure 4.5.2 shows the chromatogram of the same extract spiked with moniliformin standard. The observation of a single peak in the chromato graphic position of moniliformin in the spiked extract (Figure 4.5.2) provided evidence for the presence of moniliformin. 2 Moniliformin Moniliformin 0 2 6 8 10 0 2 4 6 8 10 Time FIG. 4.5 (1) HPLC-UV chromatogram of a sample extract (Bizana #11) and (2) a similar chromatogram of the same ex tract spiked with moniliformin standard 121 As detailed in section 2.8.4, extracts were also monitored at two separate wavelengths simultaneously (229 nm and 254 nm), using ~ diode array detector. The two wavelengths chosen cor responded to the two wavelength maxima observed in the UV spectrum of a moniliformin standard (Figure 3. 2). Figure. 4. 6 shows the two separate chromatograms obtained from an extract of Bizana sample number 7 (Table 4.2). A ratio of 3.5:1 (229 nm:254 nm) was obtained for the two chromatographic peaks. This ratio was identical to that observed for a similarly treated moniliformin standard and provided evidence that the compound eluting in the chromatographic position of monili- formin, in the sample extract, had the same wavelength ratio characteristics as the moniliformin standard. 120 100 Man i 1 i form in 80 i I ! :J IIt' a:: 60 E 254 nm(V~\j 40 20 229 nm \ 0 2 4 6 8 Time (min. ) FIG. 4.6 Paired ion chromatograms of a sample extract (Bizana #7) monitored at two separate wavelengths (229 nm and 254 nm) simultaneously 122 A number of sample extracts were also separated on an ion ex- change HPLC column (section 2.8.4), using the chromatographic conditions given in Table 2.9. Figure 4.7 shows the 'ion ex- change chromatogram (monitored at 229 nm) of an extract of Bizana sample number 11 (Table 4.2), showing a peak eluting at 8.1 minutes, the UV spectrum of which is $hown in the in- sert. The characteristic uv·spectrum (200-400 nm) was found to be identical to that observed for a moniliformin standard (Figure 3.2), and clearly provided conclusive evidence of the presence of moniliformin in the sample extract. 25 20 nm :J 1 5 ([ E ~ 10 254 nm 5 250 300 350 40 Wavelength ( nm) / 30 UV spectrum :J / ([ E 20 10 2 4 6 8 10 Time (min.) FIG. 4.7 Chromatogram of a sample extract (Bizana #11) sepa- rated on an ion exchange HPLC column, showing a moniliformin peak eluting after 8.1 minutes (the UV spectrum of which is shown in the insert) 123 4.3.4 Zearalenone analyses Using the chromatographic conditions detailed in Table 2.3, the presence of zearalenone was detected in ± 66% of the Bizana and ± 42% of the Kentani maize sample extracts. Figure 4.8.1 displayes the chromatogram of an extract prepared from Kentani sample number 1, whilst Figure 4.8.2 illustrates the 1 2 Zearale.none Zearalenone 0 3 6 9 0 3 6 Time FIG. 4.8 HPLC-fluorescence chromatograms of (1) a sample ex- tract (Kentani #1) and (2) the same extract spiked with authentic zearalenone 124 chromatogram of the same extract spiked with zearalenone standard. Although in this case a successful "spike" is ac companied by an increase in peak height (and area), it should be stressed that the observation of a single chromatographic peak (rather than a closely eluting secondary or shoulder peak), may be considered as confirmatory evidence for a compounds' presence. As "described in section 2.8.3, the presence of relatively high levels of zearalenone, in a number of the maize extracts was also monitored using UV detection at the three separate wavelength maxima for zearalenone (Figure ~.3). Figure 4.9 shows the three chromatograms, recorded simultaneously, of a maize sample extract (Bizana sample number 5 ... Table 4.2). 500 400 Zearalenone 300 316 nm ::J ([ E 200 2?4 nm 100 236 nm 0 2 3 4 5 6 7 8 Time Cmin. ) FIG. 4.9 HPLC-UV chromatograms of an extract of Bizana sam- ple #5 monitored at three wavelengths (236 nm, 274 and 316 nm) simultaneously 125 Table 4.2). A ratio of 5.2:2.3:1 was calculated from the three chromatograms obtained at the different wavelengths (236:274:316 nm). This ratio was found to be identical to that observed for a zearalenone standard, thereby supplying additional confirmatory evidence for the presence of zearalenone in the sample extract. The data collected on Bizana sample number 5, via the diode array detector, could also be displayed as a 3 dimensional representation (Figure 4.10). The characteristic UV spectrum for zearalenone can be seen at a retention time of 5.5 miriutes. Zeara1enone 400 2.5 Wavelength CnmJ Ttme CmtnJ FIG 4.10 Three dimensional representation of a HPLC-UV chromatogram of Bizana sample #5 126 4.3.5 Tricarballylic acid analyses As discussed in sections 3.9 - 3.10, an indication as to the possible extent of fumonisin contamination in maize samples, might be obtained by the determination of the tricarballylic acid (TCA) moiety, present in the structures of the known fumonisins. TCA was detected in each Transkeian maize sample TCA + Cll Illc Ill8. Cll ... 3 (.I ~ ~ c .2 .] .S! ·CII 8 ell fi: 0 2 4 6 8 10 12 14 16 ' 18 20 22 Time FIG. 4.11 Capillary GC-FID chromatogram of a hydrolysed~este- rified extract of Kentani sample #11, showing a major peak eluting at 18.3 minutes, corresponding to the chromatographic position of an esterified tricarballylic acid (TCA) standard 127 analysed ~sing the capillary GC-FID quantitative procedure outlined in section 2.7.3. Figure 4.11 displays the chromate- gram of a hydrolysed-esterified extract of Kentani sample num- ber 11 (Table 4.2). A major peak, corresponding to th~ chro- matographic position of a similarly derivatized TCA standard, can be seen at a retention time of 18.3 minutes. II. 1 RIC 151!8 ~SCIO< 12139 28151 TU£ Ill. I 2 IS .2 14 .I ~ .. IIIZ 141 168 Ill.I 21 .1 ~ .. 297.1 329.1 34S.2 IIIZ FIG. 4.12 (1) Capillary GC-MS chromatogram of an esterified tricarballylic acid (TCA) standard showing a major peak eluting at 17.3 minutes, (2) the mass spec- truro of the peak 128 Iri order to verify the presence of TCA in the hydrolysed- esterified maize sample extracts, several derivatives were analysed by capillary GC-MS, as outlined in section 2.8.6. Figure 4.12.1 shows the reconstructed ion chromatogram (RIC) of TCA as its iso-butyl ester. A major peak eluted at a retention time of 17.3 minutes, the mass spectrum of which is shown in Figure 4.12.2. 1 !51. • J RIC ,...... lel.t 57\Z 2 IS .1 •••• '"·' 131.1 ... z ... IR IBII.I 21 •• "·' FIG. 4.13 (1} Capillary chromatogram of a hydrolysed esteri- fied extract of Kentani sample #4, (the chromatogr aphic position of tricarballylic acid is indicated by an asterisk) (2) the mass spectrum of the peak 129 Figure 4.13.1 illustrates the RIC of a· hydrolysed-esterified extract of Kentani sample number 4, found to contain in ex -1 cess of 57 ~g g TCA (Table 4.2). A peak occurred at a re- tention time of 17.3 minutes (indicated by the asterisk). The mass spectrum of this peak is shown in Figure 4.13.2. The excellent agreement between the mass spectra of the derivatized TCA standard (Figure 4.12.2) and the peak eluting at 17.3 minutes, in the similarly derivatized maize sample hydrolysate (Figure 4.13.2), provided unequivocal evidence for the presence of TCA in the sample extract. Derivatized extracts of several other samples were similarly analysed and excellent agreement between retention times and mass spectra were obtained. In addition to the analysis of the sample extracts, by the capillary GC-FID quantitative method, attempts to quaiita tively assess the contamination of maize samples with fumoni- sin B1 , by the maleyl derivative procedure (detailed in sec tion 2.7.2) were made. Those samples having TCA levels in ex cess of 30 IJ.g g-1 (Kentani sample nu~bers 2,4,5,6,7,11 and 12), were analysed for the presence of fumonisin B1 as its maleyl derivative. Of the seven samples analysed, only one sample extract (number 11) appeared to contain a detectable level of fumonisin B . Figure 4.14.1 illustrates the-chromatogram of 1 fumomonisin B (1.25 ~g) as its maleyl derivative. Figure 1 4.14.2 shows the chromatogram of a 5 ~1 aliquot of a similarly prepared sample extract of Kentani sample number 130 11; and Figure 4.14.3 shows the chromatogram of the same extract spiked with derivatized fumonisin B1. The level of fumonisin B1 determined in the sample was found to be in the -1 order of 200 IJ.g g . However, this level (when converted) only accounted for approximately 20% of the total TCA deter mined in the sample by the capillary GC-FID procedure, which would suggest that the balance of the TCA present in this (as well as in the other samples) emanated from sources other than fumonisin B (ie possibly fumonisins B or B ) . 1 2 3 .015 1 2 3 -·s 0 M= .010 C\1 '-' Q,) tl) 0 0.= tl) ..Q,) FB1 0 .005 ~ ...Q,) Q,) "tt > ;;J ' 0 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 Time FIG. 4.14 HPLC chromatograms of (1) 1.25 IJ.g fumonisin B1 (FB ), as its maleyl derivative, (2) the maleyl 1 derivative of an extract from Kentani sample #11 and (3) the extract spiked with derivatized FB1 131 The chromatogram of the maleyl derivative of the Kentani sam- ple extract spik~d with fumonisin B (Figure 4.14.3) appeared 1 to show the presence of a shoulder peak in the·chromatograp- hie position of fumonisin B1 (retention time± 6 minutes). Poor chromatographic separation of the fumonisin B peak from 1 earlier eluting interferences necessitated the development of a superior separatory system. 0.015 1 2 3 -a = M= 0.010 5 ~ .CI) 0= .Cilc. ~ ...... FB 1 ...0 (J 0.005 ...~ ~ "tt ~ >j:) I 0 -1 0 8 16 24 32 0 8 16 24 32 0 8 16 24 32 Time FIG. 4.15 HPLC chromatograms of (1) a maleyl derivative blank, (2) a fumonisin B1 (FB1 ) standard and (3) the maleyl derivative of Kentani sample #11, sepa rated using an acetate buffer based solvent system and a Phenomenex ODS 30 HPLC column 132 Figure 4.15.1 illustrates the chromatogram of amaleyl deri vative blank separated on a Phenomenex ODS 30 reverse phase column, using an acetate buffer:methanol mobile phase (detai led in section 2.8.5). Using the chromatographic conditions outlined in Table 2.10, fumonisin B1 eluted after 24 minutes· (Figure 4.15.2). Figure 4.15.3 shows the chromatogram obtained from the maleyl derivative of an extract of Kentani sample number 11, and clearly illustrates the superior chromatograp hic separation achieved compared to that previously obtained (Figure 4.14.3). 4.4 STATISTICAL ANALYSES OF THE CHEMICAL AND MYCOLOGICAL RESULTS 4.4.1 overview The correlations between the levels of the selected Fusarium mycotoxins determined in the Transkeian maize samples, and the degree of fungal infection present in the same maize samples by the dominant Fusarium species, is given in Table 4.3. statistical correlations between the separate Fusa- rium species and the dominant toxins will be dealt with separately in sections 4.4.2 - 4.4.4. 4.4.2 Moniliformin and FUsarium subglutinans Table 4.4 illustrates the mean results of F. subglutinans and moniliforinin levels, determined in the mouldy maize sam ples collected from the two oesophageal cancer risk areas of the Transkei in 1985. 133 TABLE 4.3 CORRELATIONS BETWEEN FUSARIUM SPECIES AND TOXINS IN MOULDY TRANSKEIAN MAIZE OF THE 1985 CROP Toxin Fusarium Fusarium Fusarium moniliforme subglutinans graminearum 1 Moniliforrnin r = - 0.312 r = + 0.603 r = + 0.228 1 Zearalenone r = - 0.306 r = + 0.003 r = + 0.798 2 1 Nivalenol r = - 0.539 r = - 0 .. 171 r = + 0.855 2 1 Deoxynivalenol r = 0.512 r = - 0.176 r = + 0.650 1 2 Tricarballylic r = + 0.656 r = - 0.078 r = -· 0.602 acid 1 Significantly correlated (p <0.01) 2 Significantly negatively correlated (p <0.05) TABLE 4.4 INCIDENCE OF FUSARIUM SUBGLUTINANS AND MONILI- FORMIN IN MAIZE FROM OESOPAHGEAL CANCER AREAS IN 1 TRANSKEI, IN 1985 2 Low Rate Area High Rate Area p F. subglutinans (%) 10.1 4.7 -1 Moniliformin (JJ.g g ) 3.53 0.78 <0.01 4 F. subglutinans : Moniliformin, r = + 0.603 (p <0.01) 1 Means based on 12 samples from each area 2 Probability factor 3 Not significant 4 Correlations based on 24 samples 134 A higher mean value of kernels infected with F. subglutinans . . was observed in the Bizana district (low oesophageal cancer rate area) than in the high rate area (Kentani- Table 4.4), although the difference was not found to be statistically sig nificant. A significantly higher mean value (p <0.01) of mo- niliformin was recorded in the Bizana district than in· the Kentani district, and the percentage kernels infected with F. subglutinans was aignificantly correlated with the moni- liformin content (r = + 0.603, p <0.01). 4.4.3 Zearalenone, nivalenol, deoxynivalenol and FUsarium graminearum Fusarium graminearum was significantly correlated with zea- ralenone (r = + 0.798; p <0.01- Table 4.3), nivalenol (r = +·0.855; p <0.01) and deoxynivalenol (r = + 0.650; p <0.01). The mean values of the percentage kernels infected with F. graminearum and the zearalenone, nivalenol and deoxynivale- nol levels determined in the maize, were higher in those sam ples from the low oesopahgeal cancer risk area (Bizana) than the corresponding samples from the Kentani district (Table 4.5). With the exception of deoxynivalenol, these results were found to be significantly different, as reflected in the probability factor (p <0.05). '135 TABLE 4.5 INCIDENCE OF FUSARIUM GRAMINEARUM, ZEARALENONE, NIVALENOL AND DEOXYNIVALENOL IN MOULDY MAIZE FROM OESOPHAGEAL CANCER AREAS IN TRANSKEI, IN 19851 Low Rate Area High Rate Area p 2 F. graminearum (%) 34.9 8.0 <0.01 -1 Zearalenone (~g g ) 1.15 0.36 <0.05 -1 Nivalenol (~g g ) 4.62 1.75 <0.05 -1 3 Deoxynivalenol (~g g ) 2.91 0.29 NS 4 F. graminearum zearalenone, r = + 0.798 (p <0.01.) 4 F. graminearum nivalenol, r = + 0.855 (p <0.01) 4 F. graminearum deoxynivalenol, r = + 0.650 (p <0.01) 1 Means based on 12 samples from each area 2 Probability factor 3 Not significant 4 Correlations based on 24 samples 4.4.4 Tricarballylic acid and FUsarium moniliforme Table 4.6 displays the mean values of the percentage of ker- nels infected with F. moniliforme and the tricarballylic acid determined in the samples, from the two oesophageal can- cer rate areas of the Transkei. 135 TABLE 4.5 INCIDENCE OF FUSARIUM GRAMINEARUM, ZEARAELNONE, NIVALENOL AND DEOXYNIVALENOL IN MOULDY MAIZE FROM 1 OESOPHAGEAL CANCER AREAS IN TRANSKEI, IN 1985 Low Rate Area High Rate Area p 2 F. graminearum (%) 34.9 8.0 <0.01 -1 Zearalenone (~g g ) 1.15 0.36 <0.05 -1 Nivalenol (~g g ) 4.62 1.75 <0.05 -1 3 Deoxynivalenol (~g g ) 2.91 0.29 NS 4 F. graminearum zearalenone, r = + 0.798 (p <0.01) 4 F. graminearum nivalenol, r = + 0.855 (p <0.01) 4 F. graminearum deoxynivalenol, r = + 0.650 (p <0.01) 1 Means based on 12 samples from each area 2 Probability factor 3 Not significant 4 Correlations based on 24 samples 4.4.4 Tricarballylic acid and Fusarium moniliforme Table 4.6 displays the mean values of the percentage of ker nels infected with F. moniliforme and the tricarballylic acid determined in the samples, from the two oesophageal can- cer rate areas of the Transkei. 136 TABLE 4.6 INCIDENCE OF FUSARIUM MONILIFORME AND TRICARBAL LYLIC ACID IN MOULDY MAIZE FROM OESOPHAGEAL CANCER 1 AREAS IN TRANSKEI, IN 1985 2 Low Rate Area High Rate Area p F. moniliforme (%) 34.5 67.7 <0.01 Tricarballylic 12.4 77.7 <0.01 . . . -1 acJ.d (IJ.g g ) 3 F. moniliforme tricarballylic acidi r = + 0.656 (p <0.01) . 1 Means based on 12 samples from each area 2 Probability factor 3 Correlations based on 24 samples Far higher mean values for both tricarballylic acid and fun- gal infection with F. moniliforme, were recorded in the samples from the Kentani district (a high oesophageal cancer rate area) than in the corresponding low rate area (Bizana). ~hese differences were found to be statistically significant (p <0.01- Table 4.6), and a· significant correlation was found between the incidence of F. moniliforme and the levels of tricarballylic acid (r = 0.656; p <0.01- Table 4.3). 4.5 SUMMARY OF STATISTICAL ANALYSES From previous studies involving the screening of fungal cul- 137 tures prepared from isolates taken from Transkeian maize sam ples, moniliformin has only been found to be produced by F. subglutinans. Similarly, zearalenone, nivalenol and deoxy nivalenol have only been found to be produced by F. grami- nearum. Conversely, F. moniliforme isolates have never been found to produce any of these Fusarium mycotoxins. Therefore, the overall correlations as given in Table 4.3, agree with the toxin producing abilities of the Fusarium species, isolated from the series of samples, and clearly de monstrate the excellent compatibility that exists between the mycological enumeration and the quantitative analytical tech niques used in this investigation. Significantly negative correlations were found between the incidence of F. moniliforme and the levels of nivalenol and deoxynivalenol (r = - 0.539; p <0.05 and r = - 0.512; p <0.05, respectively- Table 4.3), as well as between F. graminearum and tricarballylic acid (r = 0.602; p <0.05). These results were indicative of a significantly negative correlation that was found between incidences of F. moniliforme and F. gra- minearum (r =- 0.683~ p <0.01), and imply that samples ha- ving high levels of F. moniliforme containing tricarbally lic acid, normally had low levels of infection by F. grami nearum and its metabolites, and vice versa. ~able 4.7 displays a similarly treated set of chemical and mycological results pertaining to studies carried out on 138 similar maize samples collected from the same oesophageal cancer risk areas in the Transkei, during 1979. These results have previously been presented {Thiel et al., 1982a). TABLE 4.7 CORRELATIONS BETWEEN FUSARIUM SPECIES AND TOXINS IN MOULDY TRANSKEIAN MAIZE OF THE 1979 CROPa Toxin Fusarium Fusarium Fusarium moniliforme subglutinans graminearum 1 Moniliformin r = + 0.108 r - + 0. 611 r = - 0.557 1 Zearalenone r = - 0.432 r = - 0.268 r = + 0.720 2 2 1 Nivalenol r = - 0.363 r = - 0.434 r = + 0.823 Deoxynivalenol r = - 0.013 r = - 0.031 r = + 0.;158 a Results previously presented {Thiel et al., 1982a) 1 ~ignificantly correlated (p <0.01) 2 Significantly negatively correlated (p <0.05) Although the methods pertaining to the mycological investiga- . tions carried out on the 1979 maize samples, were the same as used on the 1985 series of samples, the chemical methodo logy differed considerably. Moniliformin and zearalenone levels were determined by HPLC and TLC, respectively (Thiel et al., 1982b). Nivalenol and deoxynivalenol levels were de .termined using a packed column GC-FID technique. F. subglutinans was significantly correlated with monili- 139 formin content, while F. graminearum was similarly correla- lated with both zearalen~ne and nivalenol but not deoxyniva lenol (Table 4.7). Similar correlations were therefore ob- served in the two sets of results. Figure 4.16 illustrates the distribution of Fusarium moniliforme and tricarbally lic acid levels determined in the 1985 Transkeian maize sam- ples. Figure 4.16 demonstrates the presence of two distinct po- pulations (of results), corresponding to the samples emana- ting from the separate oesophageal cancer risk areas, of the Transkei. \1 Low oesophageal cancer risk area (Bizana) 500,000 • e High oesophageal cancer risk area (Kentani) .... -·be Ill) E. 100.000 "'tl v '(j -; %maize kernels infected with f. moniliforme (log. scale) FIG. 4.16 Distribution of Fusarium moniliforme and tricar ballylic acid in mouldy maize from high and low oesophageal cancer risk areas of the Transkei, in 1985 140 It should be remembered that had the result of Kentani sample 1 #11 (518 IJ.g g- - Table 4.2, represented in Figure 4.16 as the highest value) been withinthe range of results reported for the other Kentani samples, the correlation between F. moniliforme and tricarballylic acid would have been superior to that reported (r = + 0.656- Table 4.3). The statistical evidence, from the present and previous stu dies, indicate that the populations of both the low and high oesophageal cancer risk areas of the Transkei, have been ex posed to high levels of a number of food-borne Fusarium my cotoxins for s~veral years. The evidence concerning the presence of tricarballylic acid suggests that individuals residing in the high oesophageal cancer risk district of Kentani, are exposed to significantly higher leve~s of the potent cancer-promoting Fusarium mycotoxins, the fumoni- sins, than those residing in the Bizana district. Whilst these results are certainly a cause for concern, they do not suggest that the fumonisins are responsible for oeso phageal cancer, however they do imply that further investiga tions, concerning the role, if any, that the Fusarium my cotoxins play in the etiology of human oesophageal cancer, are warranted. ' 141 SUMMARY Fungi or moulds are members of the lower plant species that can grow either parasitically ,on plants, animals or man or saprophytically on dead organic matter. During their growth stage, numerous fungi have the ability to produce a diverse range of secondary metabolites (mycotoxins), which can be poisonous when ingested by animals or man. In several mycological surveys, Fusarium moniliforme Shel- don has been shown to be one of the three most prevalent Fusarium species isolated from mouldy home-grown maize from different oesophageal cancer areas of the Transkei, southern Africa.· The incidence of this fungus has been correlated with human oesophageal cancer rates, both in Transkei and the Lin Xian county, in the Henan province of China, which are amongst the highest oesophageal cancer risk areas of the world. Recently a group of related compounds (the fu- monisins) were isolated from a fungal culture of F. monili- forme, which itself was an isolate from a Transkeian maize sample. Fumonisin B1 (FB1 ), the major compound, was found to possess cancer-promoting abilities. It has become important to develop and implement methodology for the determination of Fusarium mycotoxins in food sam- ples, in order to assess the possible human exposure to these potentially dangerous, naturally produced, fungal metaboli- .. tes. This study involved essentially twostages. The first stage being the provision of the most suitable analytical ' 142 procedures to determine the Fusarium mycotoxins. This was achieved via the evaluation of existing or development of in novative methodology for the chromatographic determination of the Fusarium mycotoxins, in a maize substrate. The acquisi- tion and confirmation of authentic mycotoxin standards was a prerequisite for this stage of the investigation. At least two existing methods for the determination of the trichothecenes (diacetoxyscirpenol, T-2 toxin, nivalenol, deoxynivalenol), moniliformin and zearalenone were evaluated. It was necessary to develop.innovative methodology for the determination of FB1 • Due to the absence of either uv or fluorescence characteristics, the preparation of derivatives, prior to either HPLC or capillary GC analyses, ·was a prerequisite. HPLC-UV analysis of the maleyl derivative of FB1 was found to be suitable for the determination of high concentrations of FB1 (as encountered in fungal cultures), but was unsuitable for the screening of naturally contamina ted samples. Matrix effects were found to cause problems during the formation of a FB1-fluorescamine derivative com plex, prior to HPLC-fluorescence analysis. Acid hydrolysis and derivatization of FB1 , followed by capillary GC-FID analysis, resulted in the determination of the tricarballylic acid (TCA) moiety, present in the structures of the fumoni sins. During the evaluation of these FB 1 analytical techni ques, using a Transkeian maize sample, it became apparent that other TCA containing compounds (possibly other known fumonisins) were present in maize samples. It was suggested 143 that the TCA moiety might be used as an indicator of the level of total fumonisin mycotoxin contamination of maize samples. The second stage of this study involved the application of the most suitable selected Fusarium mycotoxin methods, to a series of Transkeian home-grown maize samples, collected in 1985. Excellent statistical correlations were obtained be tween the various toxin levels and the extent of Fusarium fungal contamination of the individual samples. The correla lations agreed well .with the toxin producing abilities of the dominant Fusarium species enumerated in the Transkeian maize samples, and suggested that both the analytical and my cological techniqu~s used, were suitably compatible. The most relevant statistical correlation to be generated was that be- tween the incidence of F.. moniliforme and the concentration of TCA. If as suggested, TCA can be used as an indicator to monitor the total fumonisin contamination of maize, then the evidence suggests that humans in oesophageal cancer risk areas of the Transkei in southern Africa, were exposed to the cancer-promoting F. moniliforme mycotoxins, the fumonisins. The isolation and characterization of other fumonisin com pounds and the development of suitable analytical methodology for their determination, are continuing. Futher investigation, concerning the role, if any, that the fumonisins and/or other Fusarium mycotoxins play in the.etiology of human oesopha- geal cancer in the Transkei are envisaged. 144 REFERENCES Aucock, H.W., Marasas, W.F.O., Meyer, C.J. and Chalmers, P.J. (1980), s. Afr. Vet. Assoc., 51, 163-166. Austwick, P.K.c. (1978), In: Mycotoxic fungi, Mycotoxins, My cotoxicoses An Encyclopedic Handbook. Wyllie, T.D. and Morehouse, L.G. (eds.), Vol. 3, Marcel Dekker, Inc., New York. pp 279-301. Bagneris, R.W., Gaul, J.A. and Ware, G.M. (1986), J. Assoc. Off. Anal. Chern., 69, 894-898. Bamburg, J.R., Riggs, N.V. and Strong, F.M. (1968), Tetrahe dron, 24, 3329-3339 •. Bamburg, J.R. and Strong, F.M. (1971), In: Micobial toxins: Algal and Fungal Toxins. Kadis, s., Ciegler, A. and Ajl, ·s.J. (eds.), Vol. 7, Academic Press, New York. pp 207-292. Beister, H.E., Schwarte, L.H. and Reddy, C.H. (1940), Vet. Med., 35, 636-639. Bennett, G.A., Shotwell, O.L. and Kwolek, W.F. (1985), J. Assoc. Off. Anal. Chern., 68, 958-961. Bezuidenhout, G.C., Gelderblom, W.C.A., Gerst-Allman, C.P., Horak, R.M., Marasas, W.F.O., Spiteller, G. and Vleggaar, R. (1988), J. Chern. Soc. Chern. Commun., 743-745. 145 Blau, K. and King, G. {1977), In: Handbook for Derivatives in Chromatography. Blau·, K. and King, G. ( eds) , Heyden, London. Booth, c. (1971), In: The genus Fusarium, commonwealth My- col. Inst. Kew, England. p 237. Brian, P.W.i Dawkins, A.W., Grove, J.F., Hemming, H.G., Lowe, D. and Norris, G.L.F. (1961), J. Exp. Bot., 12f 1-12. Burgess, L.W., Wearing, A.H. and Toussoun, T.A. (1975), Aust. J. Agric. Res., 26, 791-799. Butler, T. (1902), Amer. Vet. Rev., 26, 748-751. Cook, P. (1971), Br. J. Cancer, 25, 853-880. Chang, H.L. and DeVries, J.W. (1984), J. Assoc. Off. Anal. Chem., 67, 741-744. Christensen, C.M. (1979), In: Conference on Mycotoxins in Animal Feeds and Grains Related to Animal Health, Rockville, Maryland June 8, 1979., Shimoda, W. (ed.), Report on FDA/BCM- 79/139. u.s. Dept of Commerce·, National Technical Information service, Springfield, Virginia. pp 1-79. Christensen, C.M., Nelson, G.H. and Mirocha, C.J. (1965)~ Appl. Microbiol., 13, 653~659. 146 Cohen, H. and Lapointe, M. (1984), J. Assoc. Off. Anal. Chern. 67, 1105-1107. Cole, R.J. (1986), In: Modern Methods in the Analysis and Structural Elicidation of Mycotoxins. Cole, R.J. (ed.), Aca demic Press, New York. Cole R.J. and Cox, R.H. (1981), In: Handbook of Toxic Fungal Metabolites. Cole, R.J. and Cox, R.H. (eds.), Academic Press, New York. Cole, R.J., Kirskey, J.W., cutler, H.G., Doupnik, B.L. and Peckham, J.c. (1973), Science, 179, 1324-1326. Dounin, M. (1926), Phytopathology, 16, 305-308. Eppley, R. M. , Trucksess, M. W. i Nesheim, s., Thorpe, c.w. 1 Wood, G.E. and Pohlan~, A.E. (1984), J. Assoc. Off. Anal. Chern., 67, 43-45. Francis, R.G. and Burgess, L.W. (1977), Trans. Brit. Mycol. Soc., 68, 421-427. Freeman, G.G. and Morrison, R.I. (1949), Biochem. J., 44, 1-5. Gelderblom, W.C.A., Jaskiewicz, K., Marasas, W.F.O., Thiel, P.G. and Horak, R.M. (1988), Appl. Environ. Microbiol., 54, 1806-1811. 147 Gelderblom, W.C.A., Thiel, P.G., Marasas, W.F.O. and Vander Merwe, K.J. (1984), J. Agric. Food Chem., 32, 1064-1067. Godtfredson, W.O., Grove, J.F. and Tamm, C.H. (1967), Helv. Chim. Acta., 50, 1666-1669. Grove, M.D., Yates, S.G., Tallent, W.H., Ellis, J.J., Wolff, I.A., Kosuri, N.R. and Nichols, R.E. (1970), J. Agric. Food Chem., 18, 734-736. Hagler, W.M. Jr, Tyczkowsk~, K. and Hamilton, P.B. (1984), Appl. Environ. Microbial., 47, 151-154. I Hoffmann, R. W. , Bressel, u. , Gehlhaus, J. and Heuser, H. (1971), Che•. Ber., 104, 873-885. Jaskiewicz, K., Marasas, W.F.O. and Vander Walt, F.E. (1987) s. Afr. Med. J., 72, 27-30. Joffe, A.Z. (1978), In: Mycotoxic fungi, Mycotoxins, Myco toxicoses - An Encyclopedic Handbook. Wyllie, T.D. and More- house, L.G. (eds.), Vol. 2, Marcel Dekker, Inc., New York. pp 21-86. Kamimura, H., Nishijima, M., Saito, K., Takahashi, s., Ibe, ·A., Ochiai, s. and Naoi, Y. (1978), J. Food Hyg. Soc. Jpn., 19, 443-448. 148 Kamimura, H., Nishijima, M., Yasuda, K., Saito, K., Ibe, A., Nagayama, T .• , Ushiyama, H. and · Naoi, Y. ( 1981) , J. Assos. Off. Anal. Chern., 64, 1067-1073. Kosuri, N.R., Grove, M.D., Yates, S.G., Tallent, W.H., Ellis, J.J., Wolff, I.A. and Nichols, R.E. (1970), J. Amer. Vet. Med. Assoc., 157, 938-940. Kraybill, H.F. and Shimkin, M.B. (1964), In: Carcinogenesis Related to Foods. Haddow, A. and Weinhouse, s. (eds.), Vol. a, Academic Press, New York. pp 191-248. Kriek, N.P.J., Kellerman, T.S. and Marasas, W.F.O. (1981), Onderstepoort J. Vet. Res., 48, 129-131. Kriek, N.P.J., Marasas, W.F.O., Steyn, P.S., Van ~Rensbur9, S.J. and Steyn, M. (1977), Food Cosmet. Toxicol., 15, 579-587; Krouwer, J.S. and Rabinowitz, R. (1984), Clin. Chern., 30, 290-292. Kurata, H. (1978), In: Toxicology, Biochemistry and Pathology of Mycotoxins. Uraguchi, K. and Yamazaki, M. (eds.), Kodansha Ltd., Tokyo. pp 13-64. Kurata, H. and Ueno, Y. (1983), In: Toxigenic Fungi Their Toxins and Health Hazard. Kodansha Ltd, Tokyo and Elsevier Science Publishers B.V., Amsterdam. p 151. 149 Labadarios, D., Shephard, G.S., Moodie, I.M. and Botha, E. (1984), s. Afr. J. Sci., 80, 240-241. Lauren, D.R. and Greenhalgh, R. (1987), J. Assoc. Off. Anal. Chern., 70, 479-483. LeBreton, E., Frayssinet, c. and Boy, J. (1962), c. R. Acad. Sci. (Paris), 255, 784-786. Li , · M. , Lu, s. , Ji , c. , Wang, M. , Cheng, s. and Jin, c. (1979), Sci. sin., 22, 471~477. Li, M., Lu, s., .Ji, c., Wang, Y., Wang, M., Cheng, s. and Tian, G. (1980), In: Genetic and Environmental Factors in Ex perimental and Human Cancer. Gelboin, H.V., (ed.), Japan Sci. Soc. Press, Tokyo. pp 139-148. Lutsky, I., Mor, N., Yagen, B. and Joffe, A.z. (1978), Taxi col. Appl. Pharmacal., 43, 111-124. Manns, T.F. and Adams, J.F. (1923), J. Agric. Res., 23, 495. McNutt, S.H., Purwin, P. and Murray, C.J. (1928), Amer. Vet. Med. Assoc., 73, 484-492. Marasas, W.F.O. (1982), In: cancer of the Esophagus. Pfeiffer, c.J. (ed.), Vol. 1, C.R.C. Press, Boca Raton, Florida. pp. 29-40. 150 Marasas, W.F.O. (1988), In: South African Medical Research Council, Twenty Years of Growth. Brink, A.J. (ed.), Owen Bur gess, Cape Town. pp·41-45. Marasas, W. F. o. , Jaskiewicz, K. , Venter, F. s. and Van Schalk . wyk, D.J. (1988a), s. Afr. Med. J., 74, 110-114. Marasas, W.F.O., Kellerman, T.S., Gelderblom, W.C.A., Coetzer, J.A.W., Thiel, P.G. and Vander Lugt, J.J.~(1988b), Onderstepoort J. Vet. Res., 55, 197~203. Marasas, W.F.O., Nelson, P.E. ·and Toussoun, T.A. (1984), In: Toxigenic Fusarium species: Identity and Mycotoxicology. The Pennsylvania State University Press, University Park, Pennsylvania. · Marasas, W.F.O., Van Rensburg, s~J. and Mirocha, C.J. (1979), J. Agric. Food Chem., 27, 1108-1112. Marasas, W.F.O., Wehner, F.c., Van Rensburg, S.J. and van Schalkwyk, D.J. (1981), Phytopathology, 71, 792-796. Mirocha, C.J. and Pathre, s. (1973), Appl. Microbial., 26, 719-724. Morooka, N. and Tatsuno, T.· (1970), In: Toxic Micro-orga nisms. Herzberg, M., (ed.), UJNR Joint panels on micro organisms and the U.S. Department of Interior, Washington, D.C. pp 114-119. 151 Nelson, G.H., Christensen, C.M. and Mirocha, C.J. (1973), J. Amer. Vet. Med. Assoc., 163, 1276-1277. Nelson, P.E., Toussoun, T.A. and Marasas, W.F.O. (1983), In: Fusarium: an Illustrated Manual for Identification. The Pennsylvania State University Press, University Park, Penn sylvania. Official Methods of Analysis (1984), In: Association of Offi cial Analytical Chemists. Williams, s. (ed.), Vol. 14, AOAC, Arlington, Virginia. pp 477-499. Park, D.L., Miller, B.M., Hart, L.P., Yang, G., McVey, J., Page, S.W., Pestka, J. and Brown, L.H. (1989), J. Assoc. Off. Anal. Chern., 72, 326-332. Perrett, D. (1985), In: Chemistry and Biochemistry of Amino Acids. Barrett, G.C. (ed.), Chapman and Hall, London. pp 426- 461. Peters, A.T. (1904), Agric. Exp. Stn. Nebraska, 17th Ann. Rep., pp 13-22. Pinske, w. (1972), Dissertation, Technische Hochschule Aachen; cited by Scharf, H.-D. (1974), Agnew. Chem. Int. Educ., 1, 520-533. Pohland, A.E., Schuller, P.L., Steyn, P.S. and Van Egmond, H. P. (1982), Pure App. Chem., 54, 2219-2284. 152 Romer, T.R. (1983), Proc. Amer. Chern. Soc. symp., 234, 241-247. Rose, E. F. ( 1973) , J. Nat. Cancer Inst. , 51, 7-16. Rose, E.F. and McGlashan, N.D. (1975), Br. J. Cancer, 31, 197-206. Rosenthal, G.A. (1985), In: Chemistry and Biochemistry of Amino Acids. Barrett, G.C. (ed.), Chapman and Hall, London. pp 573-590. Saccardo, P.A., (1881), Fungi Italici Autographice Delinati, Patavii, 789. Schoental, R. (1984), Perspect. Biol. Med., 28, 117-120. Schuller, P.L., Van Egmond, H.P. and Stoloff., L. (1981), Pro ceedings International Symposium Mycotoxins, Cairo, Egypt, 6- 8 Sept. pp 111-129. Scott, P.M. (1983), In: Trichothecenes- Chemical, Biological and Toxicological Aspects. Ueno, Y. (ed.), Kodansha Ltd, Tokyo and Elsevier Science Publishers B.V., Amsterdam. pp 218-220. Scott, P.M ..(1987), J. Assoc. Off. Anal. Chern., 70, 276-281. Scott, P.M. (1988), J. Assoc. Off. Anal. Chern., 71, 70-76. 153 Scott, P.M. (1989), J. Assoc. Off. Anal. Chern., 72, 75-80. Scott, P.M., Kanhere, S.R. and Tarter, E.J. (1986), J. Assoc. Off. Anal. Chern., 69, 889-893. Scott, P.M., Lau, P.-Y. and Kanhere, S.R. (1981), J. Assoc. Off. Anal. Chern., 64, _1364-1371. Scott, P.M. and Lawrence, G.A. (1987), J. Assoc. Off. Anal. Chern., 70, 850-853. Scott, P.M., Nelson, K., Kanhere, S.R., Karpinski, K.F., Hay ward, s., Neish, G.A. and Teich, A.H. (1984), Appl. Environ. Microbiol., 69, 884-886. Seitz, L.M. and Bechtel, D.B. (1985), J. Agric. Food Chern~, 33, 373-377. Shank, R.c. (1978), In: Mycotoxic fungi, Mycotoxins, Mycotox icoses - An Encyclopedic Handbook. Wyllie, T.D. and Morehouse, L.G. (eds.), Vol. 3, Marcel Dekker, Inc., New York. pp 1-20. Sheldon, J.L. (1904), Agric. Exp. Stn. Nebraska, 17th Ann. Rep., pp 23-32. Shepherd, M.J. and Gilbert, J. (1986), J. Chrornatogr., 358, 415-422. / 154 Shotwell, O.L., Bennett, G.A., Stubblefield, R.D., Shannon, G.M., Kwolek, W.F. and Plattner, R.D. (1985), J. Assoc. Off. Anal. Chern., 68, 954-957 • ., Siler, D.J. and 'Gilchrist; D.G. (1982), J. Chromatogr., 238, 167-173. Snyder, w.c. and Hansen, H.N. (1940), Amer. J. Bot., 27, 64-67. Springer, J.P., Clardy, J., Cole, R.J., Kirskey, J.W., Hill, R.M., Carlson, R.M. and Is~dor, J.L. (1974), J. Amer. Chern. Soc., 92, 2267-2268. steyn, P.S. (1981), Pure Appl. Chern., 53, 891-902~ stab, M., Baldwin, R.S., Tuite, J., Andrews, F.N. and Gil- lette, K.G. (1962), Nature, 196, 1318. Sydenham, E.W., Gelderblom, w.c.A., Thiel, P.G. and Marasas, W.F.O. (1989a), J. ·Agric. Food Chern., (in press). Sydenham, E.W., Thiel, P.G. and Marasas, W.F.O. (1988), J. Agric. Food Chern., 36, 621-625. Sydenham, E.W., Thiel, P.G., Marasas, W.F.O. and Nieuwenhuis, J.J. (1989b), J. Agric. Food Chern., 37, 921-926. sydenham, E.W. Thiel, P.G., Marasas, W.F.O. and Lamprecht, S.C. (1989c), Mycotoxin Res., (in press). 155 Tanaka, T., Hasegawa, A., Matsuki, Y., Ishii, K. and Ueno, Y. (1985a), Food Add .. contam., 2, 125-137. Tanaka, T., Hasegawa, A., Matsuki, Y., Matsui, Y., Lee, U.-s. and Ueno, Y. (1985b), J. Food Hyg. Soc. Jpn., 26, 519-:-522. Thiel, P.G., Marasas, W.F.O. and Meyer, ~.J. (1982a), In: Proceedings International Conference on Mycotoxins and Phyto toxins, Vienna, Austria, 1-3 Sept. pp 126-129. Thiel, P.G., Meyer, C.J. and Marasas, W.F.O. (1982b), J. Agric. Food Chern., 30, 308-312. Tiraboschi, c. (1905), Ann. di Bot., 2, 137-168. Toussoun, T.A. and Nelson, P.E. (1976), In: A Pictorial Guide to the Identification of Fusarium species according to the Taxonomic system of Snyder and Hansen, 2nd edition. The Penn sylvania state University Press,. University Park, Pennsylva nia. Trenholm, H.L., cochrane, W.P., Cohen, H., Elliot, J.I., Farnworth, E.R., Friend, D.w., Hamilton, R.M.G., Neish, G.A. and standish, J.P. (1981), J. Am. Oil Chern. Soc., 68, 992A- 994A. 156 Trenholm, H.L., Cochrane, W.P., Cohen, H., Elliot, J.I., Farnworth, E.R., Friend, o.w., Hamilton, R.M.G., standish, J. F. and Thompson, B.K. (1983), J. Assoc. Off. Anal. Chern., 66, 92-97. Trenholm, H.L., Warner, R.M. and Fitzpatrick, D.W. (1984), J. Assoc. Off. Anal. Chern., 67, 968-972. Tsunoda, H. (~970), In: Toxic Micro-organisms. Herzberg, M. . . (ed.), UJNR Joint panels on toxic micro-organisms and the U.S. Department of the Interior, Washington, D.C. pp 143-162. Ueno, Y. (1977), Pure Appl. Chern., 49, 1737~1745. Ueno, Y. (1980), Adv. Nutr. Res., 3, 301-353. Ueno, Y. (1983), In: Trichothecenes: Chemical, Biological and Toxicological Aspects. Ueno, Y. (ed.), Kodansha Ltd, Tokyo and Elsevier Science Publishers B.V., Amsterdam. Ueno, Y., Ishii, K., Sakai, K., Kanaeda, s., Tsunoda, H., Tanaka, T. and Enomoto, M. (1972), Jpn. J. Exp. Med., 42, 187-203. Urry, W.H., Wehrmeister, H.L., Hodge, E.B. and Hidy, P.H. (1966), Tetrahedron Lett., 27, 3109-3114. Van Egmond, H.P. (1989), Food Addit. Contam., 6, 139-188. 157 Vesonder, R.F., Ciegler, A. and Jensen, A.H. (1973), Appl. Microbiol., 26, 1008-1010. Von Deckenbach, C~ (1907), Centralbl. Bakt., Abt. 1, 45, 507- 512. Wilson, B.J. (1971), In: Mycotoxins in Human Health. Purchase, I.F.H., (ed.), Macmillan, London. pp 223-229. Wilson, B.J. and Maronpot, R.R. (1971), Vet. Res., 88, 484-486. Wollenweber, H.W. and Reinking, O.A. (1935), Die Fusarien. Paul Parey, Berlin, (in German). Yang, c.s. (1980), Crincer Res., ~0, 2633-2644. Yoshizawa, T. (1983), In: Trichothecenes: Chemical Biological and Toxicological Aspects. Ueno, Y. (ed.), Kodansha Ltd., Tokyo and Elsevier Science Publishers B.V., Amsterdam. pp 195-209. Yoshizawa, T. (1984), In: Toxigenic Fungi- Thier Toxins and Health Hazards. Kurata, H. and. Ueno, Y. (eds.), Kodansha Ltd., Tokyo and Elsevier Science Publishers B.V., Amsterdam. pp 292-300. Yoshizawa, T. and Morooka, N. (1977), In: Mycotoxins in Human and Animal Health. Rodricks. J.V., Hesseltine, c.w. and Mohl man. M.A. (eds.), Pathatox Publishers, Park Forest South, Illinois. pp 309-321.